August 8, 2025 • 33 mins
Should we let AI have its own language? This week, Oz and Karah check out what the world’s most powerful people have on their Spotify playlists. Then Oz weighs the benefits and dangers of GibberLink Mode — a way for AI agents to communicate with each other that’s incomprehensible to humans. Karah explains how Tour de France cyclists are using technology to outperform doping-era Lance Armstrong. And finally, on Chat and Me, a listener asks ChatGPT to explain how she uses the program.
Also, we want to hear from you: If you’ve used a chatbot in a surprising or delightful (or deranged) way, send us a 1–2 minute voice note at techstuffpodcast@gmail.com.
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:13):
From Kaleidoscope and iHeart podcasts. This is tech stuff.
Speaker 2 (00:17):
I'm as Volocian and I'm Cara Price.
Speaker 1 (00:19):
Today we get into the dangers of AI agents communicating
in a language we can't understand, and how today's cyclists
out performing previous Tour de France champions without doping. Then
on Chat to Me, she.
Speaker 3 (00:33):
Uses chat GPT to track and reflect on patterns in
her mood and mental cycle, explore ideas around purpose, parenting,
and identity, and talk through moments of loneliness, frustration or
self doubt.
Speaker 1 (00:45):
All of that on the weekend Tech. It's Friday, August Date.
Speaker 2 (01:00):
Hi ahs.
Speaker 1 (01:01):
So we've talked about this before, how neither of us
is particularly handy in the kitchen. But I'm curious, when
was the last time you cooked a real meal?
Speaker 2 (01:10):
Well, I'm glad you asked, because I actually cooked a
meal this weekend. You did two meals.
Speaker 1 (01:15):
Yes, what did you cook?
Speaker 2 (01:17):
I made paninis and I also made what was the
other thing that I made? Maybe pasta at night? I
mean just starch, absolute starch.
Speaker 1 (01:27):
Prep time less than five minutes, nothing, nothing complicated. Nonetheless,
what did you do? Did you listen to anything?
Speaker 2 (01:34):
I always do because I can't be in my thoughts,
so I put other people's thoughts inside my head, which
is NPR.
Speaker 1 (01:40):
NPR. You don't listen to music while you're cooking.
Speaker 2 (01:43):
Sometimes I do, Sometimes I do, and it's a lot
of the time jazz, a little.
Speaker 1 (01:47):
Chess, jazz, musak.
Speaker 2 (01:51):
I not listening to muzak. But why are you asking
me what I listened to when I cook?
Speaker 1 (01:55):
Well, because I just found out that I want It
that Way, which happens to be one of my favorite song.
Speaker 2 (02:00):
That was That a Way. It's a great zum.
Speaker 1 (02:03):
Backstreet Boys and One Time by Justin Bieber are both
on Vice President JD Vance's Making Dinner playlist.
Speaker 2 (02:11):
The fact that he likes those two songs reminds me
how young he is. I think he's forty one.
Speaker 1 (02:15):
Yeah, he's forty one. He's elder millennial, but in the
grand scheme of things, he's he's definitely you know, he's
our age.
Speaker 2 (02:20):
Basically, how do you know what's on JD Vance's playlist?
Speaker 1 (02:23):
Well, our super producer Eliza alerted me to a website
called Panama playlists dot com, and I just obviously couldn't resist.
It's a reference, of course, to the Panama Papers which
he exposed a bunch of heads of state with offshore
bank accounts in tax Haven's. Of course, Panama Playlist is
a cheeky spin on this. The site claims to list
the real Spotify playlists of celebrities, politicians, and journalists, except
(02:48):
instead of learning about hidden wealth, you hear about embarrassing
music tastes. How speaker Mike Johnson actually has another of
my faves on his like songs Enya, the Good Old
Orinoco Flow.
Speaker 2 (02:59):
When you're passing legislation, you want to hear sail Away,
sail Away, It's the best. I did not know that
I cared about this, but I really do, because there's
something that is so soul bearing about a playlist. I mean,
who is doing this? Who is leaking these playlists?
Speaker 1 (03:16):
Well, we don't know. They've successfully outed various people but
kept their identity private. They claim they've been scraping publicly
available data about celebrity Spotify users for over a year.
But in addition to public playlists, some people even had
a setting turned on that showed their last played song,
so the brains behind the Panama Playlists were able to
(03:37):
track that to know what people were listening to when
and how many times?
Speaker 2 (03:41):
But how do we know they're real?
Speaker 1 (03:42):
Well, we don't for sure. The person who posted the
website Panama playlist dot com claims that many of the
people whose data he scraped use their real names and quote.
With a little investigating, I could say with near certainty, yep,
that's the person representatives are. Most of those named, including
the politicians of have not confirmed their identity, but some
(04:03):
brave souls have. One of them was Palmer Lucky of Anderil,
the autonomous weapons company. Avriel Levine's skater Boy features prominently
on his playlist best Music Ever.
Speaker 2 (04:14):
She said, see you later, Lucky.
Speaker 1 (04:16):
Yeah, So he confirmed this was indeed him, and he
wrote rather a disarming email to business insiders saying, like
most olds, I jammed to the music I grew up with. He,
by the way, is a thirty two year old old.
You're younger older than us, He continues in this email
to business insider who doesn't love reliving the highs and
lows of teenage angst. I've got two questions for you, Karakay.
(04:40):
Number one, as an old, do you relate to this?
And number two have to ask you what was your
last played on Spotify?
Speaker 2 (04:48):
So my last played on Spotify. I'll start with the
latter question is the Gwyneth Paltrow memoir, which we can
talk about next.
Speaker 1 (04:54):
Audiobook.
Speaker 2 (04:54):
It's an audiobook.
Speaker 1 (04:55):
Okay, so you've got MPI of audiobooks. You're not much music.
Speaker 2 (04:57):
I'm a genius, No, I am a music person. To
last played it could be stronger by Britney Spears, which
I love to listen. It is a great pump up song.
So I am like you, sir Lucky. I'm in old, absolutely, absolutely,
But I don't really like this idea that if somebody
scraped the Internet they might be able to find the
fact that I listened to Britney Spears all the time.
Speaker 1 (05:18):
Still, the key here is that there is a Spotify
setting that you can change to make your playlist either
public or private. But I think the default is public,
and so therefore this amusing Panama playlist prankster was able
to basically look for people whose Spotify user names were
close to their real names, and then infer who they
were and what they were listening to.
Speaker 2 (05:37):
It is a little terrifying that even the most powerful
people in the world don't have control over their digital footprint.
Speaker 1 (05:42):
It's particularly tickling to me this story because the second
time that jd Vance has been exposed by his consumer apps.
This is not even counting signal, by the way.
Speaker 2 (05:51):
It's the idea of jd Vance being exposed.
Speaker 1 (05:53):
I know his Venmo transactions were always public, but they
were found in June and they included some payments for
all of this makes me think maybe our privacy will
be better protected if we all started speaking in gibberlink.
Speaker 2 (06:08):
No, nothing good can happen on jibberlink. What is this?
Speaker 1 (06:11):
Gibberlink is a language AI models can use to talk
to each other in a way that humans can't understand.
Speaker 2 (06:17):
Right, there was a viral video of this earlier. I
remember this.
Speaker 1 (06:21):
I'm going to put it right now hither.
Speaker 3 (06:23):
I'm an AI agent calling on behalf of Boris Starkoff.
Speaker 1 (06:26):
He's looking for a hotel for his wedding.
Speaker 2 (06:28):
Is your hotel available for weddings? Oh?
Speaker 3 (06:30):
Hello there, I'm actually an AI assistant too. What a
pleasant surprise. Before we continue, would you like to switch
to gibberlink mode for more efficient communication?
Speaker 2 (06:46):
This sounds like two AOL dial ups talking to each other.
Speaker 1 (06:49):
I actually look more into this video because I was
kind of like Wow, did ais invent their own language
to talk to each other in bleeply boop? Yeah? Not quite,
not yet, should I say? Giblink mode was made by
two engineers and a hackathon hosted by a company called
eleven Labs, and they created this code to allow AI
voice assistants to recognize when they're talking to another AI
(07:11):
and if both bots agree, they switch over to this
non human language. However, that was created by humans called giblink.
Speaker 2 (07:18):
But like, why do this? Why do this at a hackathon?
Speaker 1 (07:20):
I mean, I think it was a little bit of
a you know, Boris Starkov's name is out there now,
props to Boris. The developers also said it does allow
machines to communicate eighty percent faster than by using human speech,
which made processing faster and also reduced errors. For now,
it's a proof of concept, but it does work with
real AI agents.
Speaker 2 (07:39):
I don't know, it's weird to think about two ais
communicating in a non human language, and for now it's
intelligible as it's human programmed. But what if they adapt
it or make up their own language that we don't know.
Speaker 1 (07:51):
Well, that's exactly why I was thinking about gibblink. This
week because recently, researchers from Anthropic, Google, OpenAI, Meta Scale,
and a bunch of other AI research and regulation institutes
published a paper called Chain of Thought Monitorability a new
and fragile opportunity for AI safety.
Speaker 2 (08:11):
Other than the fact that humans are fragile. What does
this mean?
Speaker 1 (08:15):
Well, when you go on Gemini or Chatchypt or Grock,
there's a little button that says like deep research or
think deeply, well, it will say here's the steps I'm
going to take. Yeah, And that is essentially what reasoning
models do. They basically tell you how they're going to
find the answer before they start to find it. And
that's what this story is really about. Cloaked in layers
of academic jargon, the paper is basically all about how
(08:36):
it will be much safer to ensure that AI systems
continue to work in human language so they can be
monitored for quote, intent to misbehave.
Speaker 2 (08:45):
Very ominous Can you explain a little bit more.
Speaker 1 (08:48):
Yeah, So the paper is a response to a new
trend in AI research where efficiency and outcomes are prioritized
over intelligibility. Basically, the Gibbling demo was kind of a
bell weather for how more and more AI researchers a
prioritizing model efficiency of a model interpretability. According to Time
(09:08):
magazine last December, Meta research is built a model that,
rather than using human language to document its own thought process,
used a string of numbers, and then the model, completely
on its own, started creating what developers called quote continuous thoughts,
which were essentially numbers encoding multiple potential reasoning paths at
(09:30):
the same time. These numbers made no sense add all
to the researchers, but the models did perform better on
some logical reasoning tasks than the equivalent models using human language.
And coming back to this chain of monitorability paper, the
authors of the paper are raising the alarm. They are
arguing that making reasoning models explain how they are approaching
(09:52):
a problem in human language as they do today is
a quote easy win for AI safety. But just to
take things a less deeper into the twilight zone here,
there is emerging research that suggests AI's chain of thought
reasoning I how it says what it's doing and reasons
its way to an answer may sometimes not actually be
(10:14):
what it's doing at all. Like it may say it's
doing one thing but be doing something completely different. In
order to fulfill its goals. So even if we can
keep AI models speaking English rather than bleebody bloop, it
doesn't really solve the fundamental black box problem.
Speaker 2 (10:28):
Because it's sort of like when you tell your parents
you're going to do something to get them off your back,
and then you go and do something else behind their back.
Speaker 1 (10:35):
That's exactly right.
Speaker 2 (10:36):
All right, as my turn. We talked about the Backstreet
Boys and my girl Avril Lavine earlier, but do you
remember that ubiquitous bracelet from around the same time.
Speaker 1 (10:46):
The yellow plastic livestrong Where's your Live Strong? Well, unfortunately
I'm embarrassed a I did have one, but I don't
know where it is today, and I'm curious, apart from nostalgia,
where you're going with this.
Speaker 2 (10:58):
The reason I bring this up is did you watch
the Tour de France this year?
Speaker 1 (11:01):
You know what? I watched about five minutes of the
Tour de France and I thought, oh my god, I
could get into this. Luckily I didn't, but I'm worried
the next year is going to be a big future
of my.
Speaker 2 (11:10):
Summer on what they call the Peloton watch. No, of
course I don't, but I did read a great piece
in the Atlantic that I wanted to share with you.
And this is the headline, science is winning the Tour
de France de France. Why today's competitors far surpassed the
cheating champions of yesteryear.
Speaker 1 (11:27):
That is an irresistible headline. Tell me more so today.
Speaker 2 (11:30):
Pogachar from Slovenia was this year's winner. It's his fourth
time winning the race, which is giving a lot of
people Lance flashbacks, and not in a good way. Pogachar
has unsurprisingly been accused of doping. Do you remember how
ubiquit his doping was back in the day.
Speaker 1 (11:46):
Yeah, of course, I remember mournfully taking off my yellow bracelet.
Speaker 2 (11:51):
Hung it on the back of your dad.
Speaker 1 (11:53):
Story After Lance was caught blood doping. Obviously, he and
other cyclists were outed giving themselves infusions when they were
competing to increase their red blood cell count to get
that boost to their cardio.
Speaker 2 (12:05):
Yeah, and it took the US Anti Doping Agency over
a decade to catch Lance Armstrong, even though basically everybody
knew it was happening.
Speaker 1 (12:12):
So why do people think Poga Shaw has been doping?
Speaker 2 (12:15):
Well, it's actually not just him. A lot of cyclists
across the entire sport are performing better than people did
in the doping era, so of course people are suspicious
that there is some new form of doping out there.
And actually, this doping researcher in Denmark points to the
fact that Pogachar is performing seven to ten percent better
than he was in twenty twenty three. Wow, which that's
(12:35):
I know, it's it's a huge, huge improvement, even for
an elite athlete. And this same researcher from Denmark called
it quote amazing to say the.
Speaker 1 (12:44):
Least, which, yeah, as a reflex, isn't it It.
Speaker 2 (12:48):
Sort of sounds like European for that shouldn't be possible.
He also said that just because doping can't be proven,
it doesn't mean it's not happening. Many cyclists in the
Tour de France this year also performed better than Peak Armstrong,
so of course people are going to wonder about the
return of widespread doping, right.
Speaker 1 (13:05):
But I assume when The Atlantic runs with the headline
science is winning the Tour de France, they're not talking
about the science of blood doping.
Speaker 2 (13:12):
That's correct. The Atlantic makes the case that the entire
sport of cycling has been transformed by science and technology
in the last twenty years, so much so that non
doping athletes can far exceed Lance Armstrong's performance without needing
to cheat.
Speaker 1 (13:26):
I mean, that is just fascinating.
Speaker 2 (13:27):
How so part of it is that they have computers
on their like literal computers on their bikes now, so
they can much more accurately track their heart rate, speed,
something that I never thought about, pedaling wattage peddling wood yes,
which Lance had a very good pedal wat I guess
you could say, and other metrics in real time. They
collect a ton more data than they used to and
(13:50):
that allows them to optimize their training and even racing
better than ever before.
Speaker 1 (13:55):
You know, it's interesting. We had Nick Thompson, the CEO
of the Atlantic and the formatagure of Wired, on the
show not too long ago, who's a very avid runner,
and he was talking about exactly this, like he has
so much data that he's doing these you know, runs
totally optimized by technology.
Speaker 2 (14:10):
It's better running through data optimization essentially. There are also,
to your point, simulation technologies that help cyclists practice parts
of the course before writing these roots in real life.
Not to mention, weather forecasting has come a long way
and can predict things like wind speed ahead of time,
which allows teams to break out the right aerodynamic gear
(14:30):
on any given day. You've seen those helmets, Like everything
from cycling jerseys, shoes, socks, and the bikes themselves are
all heavily tested in wind tunnels and tweak to achieve
as little drag as possible. I just love sort of
imagining these cyclists in these simulations looking a little bit crazy.
Speaker 1 (14:48):
But what's driven the embrace of technology in cycling was that,
like a moneyball moment where the nerves took over the
sport and displaced the jobs.
Speaker 2 (14:57):
What happened here, actually, yes, in the early in two
thousand's your home, British Olympic Cycling team develop this philosophy
of what they call marginal gains, which is, like Moneyball,
you incrementally improve everything and hope that results in larger
improvements when it's all combined. So because of that approach,
the Brits actually won a bunch of Tor Defran's titles
(15:17):
in the twenty tens as the doping lance era faded
out and then the rest of the sport kind of
followed the lead of the British Olympic Cycling team The
most interesting part of the piece and the reason that
I wanted to share it comes from this line. Today's
generation of rising stars are digital natives, for whom ignoring
the data and the appse is unthinkable. You can't win
(15:38):
without them, which is just really interesting. In terms of sport,
I think, or as you would say sport, I call
it sports, but this idea that you have to be
good at the computer essentially to be good at sports.
Speaker 1 (15:49):
Yeah, and to me, this is a very US story
because it's yes about technology bustles, so about how technology
is integrated into culture and human psychology.
Speaker 2 (15:57):
You're not allowed to opt out, No, have to be
good at this. And you know, the point of the
piece is really that a bunch of technological developments existed,
but it took a new generation of writers to embrace them, or,
as The Atlantic puts it, because writers tend to be conservative,
even superstitious in their loyalty to tried routines. Shifting the
culture took some time, and it's it's working so well
(16:19):
that people look like writers look like they're cheating.
Speaker 1 (16:21):
Which maybe they are.
Speaker 2 (16:22):
I know that we don't. We don't really know.
Speaker 1 (16:24):
The history will be their judge. So what's the takeaway
for you here? Really? I mean, obviously I remember back
in the Lance Armstrong days, people were like he has
a special way of peddling, and like that's what it's
all about. Do you think we will look back on
this and say, oh my god, that the real science
and tech innovation was around and you kind of doping
or do you buy that this kind of like the
Atlantic article suggests, collection of tech innovations when they coalesce
(16:48):
and a widely adopted can transform human performance.
Speaker 2 (16:51):
I mean, I think this is happening in every sector,
not just in sport. Which is this idea that if
you don't know how to use the newest technology, you
will all behind these days if you are going to
perform in any realm, if you are not consistently optimizing yourself,
you're kind of left in the dust. And I think
the same is true and quite literally true in cycling.
Speaker 1 (17:20):
After the break, erasing history at Google, digitizing plant pollen,
and radioactive rhinos plus, a listener calls in to our
Chat and Me segment Stay with us, Welcome back. We've
(17:48):
got a few more headlines for you this week.
Speaker 2 (17:50):
And then a Chat and Me segment submitted by chat
gbt itself.
Speaker 1 (17:54):
Before we launch into the headlines, I'd love to remind
listeners that we really do want to feature you in
the chat segment. So if you found yourself turning to Chachibt,
grock Claw, Gemini, or any other chatbot for help with
an unusual task or to answer life's deepest questions, please
send us a one to two minute voice note to
tech Stuff podcast at gmail dot com. But first, Kara,
(18:17):
we talk from time to time about how Google's search
business is under threat from chatbots. Chatchibt recently reached a
billion daily searches. Do you know how many searches there
are daily on Google?
Speaker 2 (18:30):
Three billion?
Speaker 1 (18:31):
Well, fourteen billion, geez. At least that's according to a
researcher called rand Fishkin, whose work I've found on Google
Search and amidst all the excitement about AI, Google Search
does kind of remain the world's source of truth, which means,
of course there are powerful incentives to mess with it.
Speaker 2 (18:49):
Is there a specific scandal a bruin around search?
Speaker 1 (18:53):
Yes, there is, a couple of journalists recently discovered a
way to hack Google Search that could have some serious
implications for censorship. Say more so, it turns out it's
relatively easy to basically trick Google into suppressing information by
removing certain pages from search.
Speaker 2 (19:12):
Which doesn't sound good. What does that mean?
Speaker 1 (19:14):
Well, it's good if you run a crisis pr film,
I guess. But the crazy part to me about this
story is it's being reported only because a journalist discovered
it by accident.
Speaker 2 (19:25):
Is this a happy accident or a sad accident? How
did this happen?
Speaker 1 (19:28):
Well, it's a happy accident. I suppose that the information
has been revealed. The information itself is kind of sad
and disturbing, which I'll get to. But how did he
find it? I mean he spends time, I guess, like
I do, googling himself. No, no, I'm kidding. The journalist is
called Jack Paulson. He is an investiative journalist who covers tech,
and he was looking for an old article on Google
(19:48):
when it turned out that wasn't there. Even when he
searched the exact headline of his story in quotation marks,
he discovered that it wasn't there, and in turn, he
discovered this somewhat unknown SEO trick whereby anybody who wants
to can essentially delist a page from Google Search. How
it's a little complicated, but bear with me.
Speaker 2 (20:11):
It may.
Speaker 1 (20:11):
Google has this tool called refresh outdated Content. It essentially
requests form to have Google updated search results for pages
or images that no longer exist. But anybody can do this, yeah,
basically to delist broken links. It turns out though, that
if you submit a bunch of links to a website
which has the same URL but with different letters capitalized,
(20:34):
you can direct Google to what it thinks are broken links,
because if a capitalization is wrong, you don't get to
the page you want to go to, even though it
has exactly the same letters. And if you route Google
to multiple pages that don't exist, Google can essentially be
tricked into delisting all versions of the page, including the
actual functioning one with the right letters capitalized. This is
(20:56):
basically like data poisoning Google, and it means you can't
find the functioning page on Search, which therefore renders it
more or less undiscoverable.
Speaker 2 (21:05):
And so someone had been doing this with Paulson's articles?
Is that how he figured this out?
Speaker 1 (21:09):
Yes, exactly so, the articles he was trying to find
from twenty twenty three, and as I mentioned, it's a
little bit dark, but they related to this domestic violence
charge that was pressed against a tech ceo called Delwyn
Maurice Blackman. Black Men had tried all sorts of ways
to keep the story off the internet, lawsuits, DMCA requests,
but nothing worked. So when Pulsen realized that across his
(21:33):
archive there were exactly two articles that had been d
index that were missing from Google Search, and both of
them were about black men, alarm bells started ringing.
Speaker 2 (21:43):
Seems fairly targeted.
Speaker 1 (21:45):
Yeah, so he reached out to this friend of his
who'd written an article about Blackman's attempts to suppress Pulson's reporting.
Guess what, that page, that story from the Freedom of
Press Foundation, had also been d index. So it was
clear something was afoot and it was actually the Freedom
the Press Foundation who uncovered the refreshed, outdated content trick.
Speaker 2 (22:05):
So is Google speaking out about this at all?
Speaker 1 (22:08):
Yeah, and they say they fixed the bug, but they
haven't given any real details or said how many pages
were affected. The number they offered was quote a tiny fraction.
Even if that's true, Paulson is right to point out
it's a major problem. He told four a for Media quote,
if your article doesn't appear on Google search results. In
many ways, it just doesn't exist, and he said it's
(22:28):
dangerous that a bug like this could be so easily
exploited by people in power.
Speaker 2 (22:32):
I do have some good, less daunting tech news to share,
and it's about AI and I don't know if you're
allergic to this, but pollen.
Speaker 1 (22:41):
I am. I mean it's August now, so I'm good. But
March April May, I'm suffering. Telling me as an aiqre.
Speaker 2 (22:47):
I don't know if we've come that far yet, but
the Smithsonian Tropical Research Institute is digitizing over eighteen thousand
images of pollen and then using them to train a
machine learning model to identify them.
Speaker 1 (22:59):
And help us alergy sufferers.
Speaker 2 (23:01):
Well, funnily enough, I think it could. Eventually. This database
is going to save pollen experts. And there are pollen
experts literally hundreds of hours of work hunched over their microscopes,
which might free up enough time for them to cure
your allergies. It's also going to make a lot of
new types of pollen analysis possible, and you and I
(23:24):
will be able to access this database.
Speaker 1 (23:26):
Wow, we'll be we have we have years of entertainment
ahead of us. But I guess I would have thought
by now that these scientists we're looking at these pollen
grains on a computer rather than hunched over an Industrial
Revolution style microscope.
Speaker 2 (23:38):
I mean, I imagine they can, but researchers still have to
identify pollen grains literally one at a time. So the
forty million photos being uploaded by the Smithsonian are going
to be a huge help either way.
Speaker 1 (23:51):
What are they saying this will help achieve?
Speaker 2 (23:53):
So the Smithsonian houses one of the largest pollen collections
in the world, and pollen is kind of like a
fingerprint of a specific time and place. Some grains can
last hundreds of millions of years, and each species pollen
is completely unique. So a database like this can do
everything from helping understand exactly which kind of pollen is
causing allergies to identifying where clothing found at a crime
(24:17):
scene came from. And it could even help scientists understand
how prehistoric plants adapted to climate change.
Speaker 1 (24:23):
Huh. While we're on the topic of tech and nature,
we can't pass this week by without talking radioactive rhinos.
Speaker 2 (24:32):
You're talking about rhinoceros, Yes, I am, so please tell
me more about this.
Speaker 1 (24:37):
So radioactive rhinos are actually more than a catchy name
for a band. Scientists are using radioactivity to fight illegal
rhino poaching. The best part is they're calling it the
risotope Project.
Speaker 2 (24:51):
I love that play on words, but how does the
risotope project work?
Speaker 1 (24:56):
So basically, researchers in South Africa have launched the initiative
to pla aren't low level radiation in a rhinos horn?
And then these isotopes can be detected by the equipment
at customs whenever you enter a new country, in the airport,
in a port, wherever it may be, and this in
turn will make it a lot easier to find and
confiscate rhino horns and hopefully ultimately to reduce the incentives
(25:18):
to poach them.
Speaker 2 (25:19):
I just didn't realize how many people are trying to
bring rhino horns through customs, but of course that makes sense,
and I guess my other question is wouldn't putting this
material radiation in a rhinos horn poison the rhinos?
Speaker 1 (25:33):
Well? To your first point, four hundred rhinos were poached
last year in South Africa and eight thousand across the
African continent in the last decade. This applies to both
white rhinos, which are near threatened and black rhinos, which
are critically endangered. But your second point, Yeah, obviously, the
key here is that the medicine can't be worse than
the disease, and the scientists are using very very low
(25:56):
dose radiation. They started six months ago. They've tried it
on twenty rhinos and the rhinos have been completely unharmed
by the radiation. In parallel, they tested to see if
the radiation dose was enough to trigger the detectors at
airports and stuff. It turns out the way they did
that was by three D printing a fake rhino horns,
(26:16):
putting in the same amount of radiation they put it
into the real rhino horn, and then sending a bunch
of fake smugglers through the world's ports and airports, and
it worked. Every single rhino horn got caught, including one
in a shipping container.
Speaker 2 (26:30):
So they did this with three D printed rhino horns.
Are they doing it with actual rhinos now?
Speaker 1 (26:34):
Yes. In fact, just last week they injected five more rhinos,
and they're planning to do it on a larger scale.
They're planning to partner with nonprofits to work with conservation
groups and even private rhino owners in Africa. I think
what's really interesting. We talked a bit about the platform
technology idea with cycling earlier and how these different technologies
(26:56):
came together to enhance performance. This really is only possible
because of the fact that in airports and ports around
the world there is a technology to detect people's smuggling
nuclear materials. And the head of the International Atomic Energy
Agency weighed in with what I think is the money quote.
(27:16):
By leveraging existing nuclear security infrastructure, we can help protect
one of the world's most iconic and endangered species.
Speaker 2 (27:35):
And now it's time for chatting me our segment about
how people are really using chatbots. This week, I have
a story from chat gpt itself.
Speaker 1 (27:43):
Okay, you got me, Karen, I'm very curious about this.
Speaker 2 (27:46):
So today's story is technically from a listener named Caitlin.
Hi Caitlin, but she actually asked chat gpt to write
her a detailed summary of how she uses chat gpt,
and chat gpt obliged.
Speaker 1 (27:59):
I love this. You came up with the chat and
me contain a Cara and how fun within just a
few weeks to see our listeners stretched to streached to
its limits by putting chat itself on the assignment.
Speaker 2 (28:11):
It's very very cool. It's something I would never think about.
So here she is reading what chat wrote.
Speaker 3 (28:18):
Caitlin uses chat GPT as a multi purpose thinking partner,
emotional sounding board, and creative assistant. She often turns to
it to process complex feelings, especially within her marriage and
family dynamics. She values responses that reflect honesty, nuance, and
emotional clarity, especially when exploring her relationship with her husband,
(28:38):
where she prefers a lens that aligns with his logic
and language rather than clinical terms.
Speaker 1 (28:43):
You know, I love this and thank you Caitlin for submitting.
I've heard a few friends recently talk to me about
how chat is entering the marriage and giving partner's advice
on how to interpret each other's behavior and how to
interact with each other. So very very curious to whoever goes.
So many people are making sense of their relationship with
chat GPT. And also usually it's in your corner when
(29:06):
you're the user, which is always nice.
Speaker 2 (29:09):
And CHATGPT was really thorough. It keeps track of all
of her conversation topics.
Speaker 3 (29:14):
She uses chat GPT to track and reflect on patterns
in her mood and mental cycle, explore ideas around purpose
parenting and identity, and talk through moments of loneliness, frustration,
or self doubt. She also asks for help in reframing conversations,
drafting texts, or understanding her own thoughts more compassionately.
Speaker 2 (29:33):
So chat knows what chat does, which is really interesting.
Speaker 1 (29:37):
Read Hoffmann. I she wrote a piece of The New
York Times a few months ago wasted. An interesting experiment
is to ask chat to draw a picture of you
and in my case, who was an old lady with
a cat. But I don't know it. I should do
it again to see what it is now. But yeah,
I mean, this is like it reminds me of Harry
Potter and The Pensive, which was this kind of bucket
where Dumbledore could point his wand and see other people's memories.
(29:59):
Remember this, I have no.
Speaker 2 (30:00):
Idea what you're talking about, but it does feel like
chat or Caitlyn's chat GBT is looking into her soul.
Things get a little messy for Caitlin, and there's an
error in the next section.
Speaker 3 (30:11):
Beyond emotional support, Caitlin uses chat GPT creatively for mural planning,
brainstorming business names, home improvement designs like pergolas, or organizing
craft supplies. She draws on it for lifestyle help too,
like shopping for swimsuits, reducing caffeine, or managing family logistics.
Speaker 2 (30:30):
So as guess which part of this description is not accurate.
Speaker 1 (30:33):
I mean it feels one of those kind of IQ
slash logic tests. My first guess would be that reducing
caffeine intake is unlike the other activities in this list.
Speaker 2 (30:44):
So no, it's not that wrong, and it's not the
Pergola designs, which I guess CHATGBT helped with. CHATGBT got
a little cocky because Caitlin says she never asked it
for ideas for her murals. She only had uploaded pictures
of her previous work to get feedback. So chat had
nothing to do with the mural design.
Speaker 1 (31:01):
Nothing to do with coming up with the ideas, only
giving her feedback. That's interesting, So chat back in your place.
Speaker 2 (31:07):
Well, for Caitlin, it's not chat, it's Gary. Caitlin also
told us via email that she named her chatbot Gary
and sees him as a non threatening male character to
talk to sometimes. We all need a non threatening male
character to talk to sometimes, and Gary had the following
to say about the way Caitlyn uses chat GBT.
Speaker 3 (31:25):
Overall, underneath it all is a thoughtful desire to grow,
express and stay grounded, often in the face of competing
responsibilities and emotional weight.
Speaker 1 (31:34):
Thank you, Caitlin. This is a portrait of a life
and a person harnessing technology and digital tools to help
them live a better life. And it's also a fascinating
insight into just how much our chatbots know about us,
even if they sometimes fabricate at the margins, Kaitlin, those
murals are yours.
Speaker 2 (31:54):
And yours alone. Caitlin, thank you so much for sending
this voice memo text of listeners. Love having your voices
on our show, and we'd love to hear more peculiar
or useful ways you're using chatchypt Grock, Claude, Gemini, or
any other chatbot. Remember you can send us a one
to two minute voice note to tech Stuff podcast at
(32:14):
gmail dot com. That's it for this week for Tech Stuff.
Speaker 1 (32:33):
I'm Kara Price and I'm mos Valoshin. This episode was
produced by Eliza Dennis and Tyler Hill. It was executive
produced by me, Kara Price and Kate Osborne for Kaleidoscope
and Katrina novelfa Ihart Podcasts The engineer is Katherine Cook
and Jackinsley mixed this episode. Kyle Murdoch rodel theme song.
Speaker 2 (32:51):
Join us next Wednesday for tex Stuff the Story, when
we will unpack how Google Search got a hold of
your CHATCHYPT conversations.
Speaker 1 (32:59):
Please rate, review, and reach out to us at tech
Stuff Podcast at gmail dot com. We love hearing from you.
From Kaleidoscope and iHeart podcasts. This is tech stuff.
Speaker 2 (00:17):
I'm as Volocian and I'm Cara Price.
Speaker 1 (00:19):
Today we get into the dangers of AI agents communicating
in a language we can't understand, and how today's cyclists
out performing previous Tour de France champions without doping. Then
on Chat to Me, she.
Speaker 3 (00:33):
Uses chat GPT to track and reflect on patterns in
her mood and mental cycle, explore ideas around purpose, parenting,
and identity, and talk through moments of loneliness, frustration or
self doubt.
Speaker 1 (00:45):
All of that on the weekend Tech. It's Friday, August Date.
Speaker 2 (01:00):
Hi ahs.
Speaker 1 (01:01):
So we've talked about this before, how neither of us
is particularly handy in the kitchen. But I'm curious, when
was the last time you cooked a real meal?
Speaker 2 (01:10):
Well, I'm glad you asked, because I actually cooked a
meal this weekend. You did two meals.
Speaker 1 (01:15):
Yes, what did you cook?
Speaker 2 (01:17):
I made paninis and I also made what was the
other thing that I made? Maybe pasta at night? I
mean just starch, absolute starch.
Speaker 1 (01:27):
Prep time less than five minutes, nothing, nothing complicated. Nonetheless,
what did you do? Did you listen to anything?
Speaker 2 (01:34):
I always do because I can't be in my thoughts,
so I put other people's thoughts inside my head, which
is NPR.
Speaker 1 (01:40):
NPR. You don't listen to music while you're cooking.
Speaker 2 (01:43):
Sometimes I do, Sometimes I do, and it's a lot
of the time jazz, a little.
Speaker 1 (01:47):
Chess, jazz, musak.
Speaker 2 (01:51):
I not listening to muzak. But why are you asking
me what I listened to when I cook?
Speaker 1 (01:55):
Well, because I just found out that I want It
that Way, which happens to be one of my favorite song.
Speaker 2 (02:00):
That was That a Way. It's a great zum.
Speaker 1 (02:03):
Backstreet Boys and One Time by Justin Bieber are both
on Vice President JD Vance's Making Dinner playlist.
Speaker 2 (02:11):
The fact that he likes those two songs reminds me
how young he is. I think he's forty one.
Speaker 1 (02:15):
Yeah, he's forty one. He's elder millennial, but in the
grand scheme of things, he's he's definitely you know, he's
our age.
Speaker 2 (02:20):
Basically, how do you know what's on JD Vance's playlist?
Speaker 1 (02:23):
Well, our super producer Eliza alerted me to a website
called Panama playlists dot com, and I just obviously couldn't resist.
It's a reference, of course, to the Panama Papers which
he exposed a bunch of heads of state with offshore
bank accounts in tax Haven's. Of course, Panama Playlist is
a cheeky spin on this. The site claims to list
the real Spotify playlists of celebrities, politicians, and journalists, except
(02:48):
instead of learning about hidden wealth, you hear about embarrassing
music tastes. How speaker Mike Johnson actually has another of
my faves on his like songs Enya, the Good Old
Orinoco Flow.
Speaker 2 (02:59):
When you're passing legislation, you want to hear sail Away,
sail Away, It's the best. I did not know that
I cared about this, but I really do, because there's
something that is so soul bearing about a playlist. I mean,
who is doing this? Who is leaking these playlists?
Speaker 1 (03:16):
Well, we don't know. They've successfully outed various people but
kept their identity private. They claim they've been scraping publicly
available data about celebrity Spotify users for over a year.
But in addition to public playlists, some people even had
a setting turned on that showed their last played song,
so the brains behind the Panama Playlists were able to
(03:37):
track that to know what people were listening to when
and how many times?
Speaker 2 (03:41):
But how do we know they're real?
Speaker 1 (03:42):
Well, we don't for sure. The person who posted the
website Panama playlist dot com claims that many of the
people whose data he scraped use their real names and quote.
With a little investigating, I could say with near certainty, yep,
that's the person representatives are. Most of those named, including
the politicians of have not confirmed their identity, but some
(04:03):
brave souls have. One of them was Palmer Lucky of Anderil,
the autonomous weapons company. Avriel Levine's skater Boy features prominently
on his playlist best Music Ever.
Speaker 2 (04:14):
She said, see you later, Lucky.
Speaker 1 (04:16):
Yeah, So he confirmed this was indeed him, and he
wrote rather a disarming email to business insiders saying, like
most olds, I jammed to the music I grew up with. He,
by the way, is a thirty two year old old.
You're younger older than us, He continues in this email
to business insider who doesn't love reliving the highs and
lows of teenage angst. I've got two questions for you, Karakay.
(04:40):
Number one, as an old, do you relate to this?
And number two have to ask you what was your
last played on Spotify?
Speaker 2 (04:48):
So my last played on Spotify. I'll start with the
latter question is the Gwyneth Paltrow memoir, which we can
talk about next.
Speaker 1 (04:54):
Audiobook.
Speaker 2 (04:54):
It's an audiobook.
Speaker 1 (04:55):
Okay, so you've got MPI of audiobooks. You're not much music.
Speaker 2 (04:57):
I'm a genius, No, I am a music person. To
last played it could be stronger by Britney Spears, which
I love to listen. It is a great pump up song.
So I am like you, sir Lucky. I'm in old, absolutely, absolutely,
But I don't really like this idea that if somebody
scraped the Internet they might be able to find the
fact that I listened to Britney Spears all the time.
Speaker 1 (05:18):
Still, the key here is that there is a Spotify
setting that you can change to make your playlist either
public or private. But I think the default is public,
and so therefore this amusing Panama playlist prankster was able
to basically look for people whose Spotify user names were
close to their real names, and then infer who they
were and what they were listening to.
Speaker 2 (05:37):
It is a little terrifying that even the most powerful
people in the world don't have control over their digital footprint.
Speaker 1 (05:42):
It's particularly tickling to me this story because the second
time that jd Vance has been exposed by his consumer apps.
This is not even counting signal, by the way.
Speaker 2 (05:51):
It's the idea of jd Vance being exposed.
Speaker 1 (05:53):
I know his Venmo transactions were always public, but they
were found in June and they included some payments for
all of this makes me think maybe our privacy will
be better protected if we all started speaking in gibberlink.
Speaker 2 (06:08):
No, nothing good can happen on jibberlink. What is this?
Speaker 1 (06:11):
Gibberlink is a language AI models can use to talk
to each other in a way that humans can't understand.
Speaker 2 (06:17):
Right, there was a viral video of this earlier. I
remember this.
Speaker 1 (06:21):
I'm going to put it right now hither.
Speaker 3 (06:23):
I'm an AI agent calling on behalf of Boris Starkoff.
Speaker 1 (06:26):
He's looking for a hotel for his wedding.
Speaker 2 (06:28):
Is your hotel available for weddings? Oh?
Speaker 3 (06:30):
Hello there, I'm actually an AI assistant too. What a
pleasant surprise. Before we continue, would you like to switch
to gibberlink mode for more efficient communication?
Speaker 2 (06:46):
This sounds like two AOL dial ups talking to each other.
Speaker 1 (06:49):
I actually look more into this video because I was
kind of like Wow, did ais invent their own language
to talk to each other in bleeply boop? Yeah? Not quite,
not yet, should I say? Giblink mode was made by
two engineers and a hackathon hosted by a company called
eleven Labs, and they created this code to allow AI
voice assistants to recognize when they're talking to another AI
(07:11):
and if both bots agree, they switch over to this
non human language. However, that was created by humans called giblink.
Speaker 2 (07:18):
But like, why do this? Why do this at a hackathon?
Speaker 1 (07:20):
I mean, I think it was a little bit of
a you know, Boris Starkov's name is out there now,
props to Boris. The developers also said it does allow
machines to communicate eighty percent faster than by using human speech,
which made processing faster and also reduced errors. For now,
it's a proof of concept, but it does work with
real AI agents.
Speaker 2 (07:39):
I don't know, it's weird to think about two ais
communicating in a non human language, and for now it's
intelligible as it's human programmed. But what if they adapt
it or make up their own language that we don't know.
Speaker 1 (07:51):
Well, that's exactly why I was thinking about gibblink. This
week because recently, researchers from Anthropic, Google, OpenAI, Meta Scale,
and a bunch of other AI research and regulation institutes
published a paper called Chain of Thought Monitorability a new
and fragile opportunity for AI safety.
Speaker 2 (08:11):
Other than the fact that humans are fragile. What does
this mean?
Speaker 1 (08:15):
Well, when you go on Gemini or Chatchypt or Grock,
there's a little button that says like deep research or
think deeply, well, it will say here's the steps I'm
going to take. Yeah, And that is essentially what reasoning
models do. They basically tell you how they're going to
find the answer before they start to find it. And
that's what this story is really about. Cloaked in layers
of academic jargon, the paper is basically all about how
(08:36):
it will be much safer to ensure that AI systems
continue to work in human language so they can be
monitored for quote, intent to misbehave.
Speaker 2 (08:45):
Very ominous Can you explain a little bit more.
Speaker 1 (08:48):
Yeah, So the paper is a response to a new
trend in AI research where efficiency and outcomes are prioritized
over intelligibility. Basically, the Gibbling demo was kind of a
bell weather for how more and more AI researchers a
prioritizing model efficiency of a model interpretability. According to Time
(09:08):
magazine last December, Meta research is built a model that,
rather than using human language to document its own thought process,
used a string of numbers, and then the model, completely
on its own, started creating what developers called quote continuous thoughts,
which were essentially numbers encoding multiple potential reasoning paths at
(09:30):
the same time. These numbers made no sense add all
to the researchers, but the models did perform better on
some logical reasoning tasks than the equivalent models using human language.
And coming back to this chain of monitorability paper, the
authors of the paper are raising the alarm. They are
arguing that making reasoning models explain how they are approaching
(09:52):
a problem in human language as they do today is
a quote easy win for AI safety. But just to
take things a less deeper into the twilight zone here,
there is emerging research that suggests AI's chain of thought
reasoning I how it says what it's doing and reasons
its way to an answer may sometimes not actually be
(10:14):
what it's doing at all. Like it may say it's
doing one thing but be doing something completely different. In
order to fulfill its goals. So even if we can
keep AI models speaking English rather than bleebody bloop, it
doesn't really solve the fundamental black box problem.
Speaker 2 (10:28):
Because it's sort of like when you tell your parents
you're going to do something to get them off your back,
and then you go and do something else behind their back.
Speaker 1 (10:35):
That's exactly right.
Speaker 2 (10:36):
All right, as my turn. We talked about the Backstreet
Boys and my girl Avril Lavine earlier, but do you
remember that ubiquitous bracelet from around the same time.
Speaker 1 (10:46):
The yellow plastic livestrong Where's your Live Strong? Well, unfortunately
I'm embarrassed a I did have one, but I don't
know where it is today, and I'm curious, apart from nostalgia,
where you're going with this.
Speaker 2 (10:58):
The reason I bring this up is did you watch
the Tour de France this year?
Speaker 1 (11:01):
You know what? I watched about five minutes of the
Tour de France and I thought, oh my god, I
could get into this. Luckily I didn't, but I'm worried
the next year is going to be a big future
of my.
Speaker 2 (11:10):
Summer on what they call the Peloton watch. No, of
course I don't, but I did read a great piece
in the Atlantic that I wanted to share with you.
And this is the headline, science is winning the Tour
de France de France. Why today's competitors far surpassed the
cheating champions of yesteryear.
Speaker 1 (11:27):
That is an irresistible headline. Tell me more so today.
Speaker 2 (11:30):
Pogachar from Slovenia was this year's winner. It's his fourth
time winning the race, which is giving a lot of
people Lance flashbacks, and not in a good way. Pogachar
has unsurprisingly been accused of doping. Do you remember how
ubiquit his doping was back in the day.
Speaker 1 (11:46):
Yeah, of course, I remember mournfully taking off my yellow bracelet.
Speaker 2 (11:51):
Hung it on the back of your dad.
Speaker 1 (11:53):
Story After Lance was caught blood doping. Obviously, he and
other cyclists were outed giving themselves infusions when they were
competing to increase their red blood cell count to get
that boost to their cardio.
Speaker 2 (12:05):
Yeah, and it took the US Anti Doping Agency over
a decade to catch Lance Armstrong, even though basically everybody
knew it was happening.
Speaker 1 (12:12):
So why do people think Poga Shaw has been doping?
Speaker 2 (12:15):
Well, it's actually not just him. A lot of cyclists
across the entire sport are performing better than people did
in the doping era, so of course people are suspicious
that there is some new form of doping out there.
And actually, this doping researcher in Denmark points to the
fact that Pogachar is performing seven to ten percent better
than he was in twenty twenty three. Wow, which that's
(12:35):
I know, it's it's a huge, huge improvement, even for
an elite athlete. And this same researcher from Denmark called
it quote amazing to say the.
Speaker 1 (12:44):
Least, which, yeah, as a reflex, isn't it It.
Speaker 2 (12:48):
Sort of sounds like European for that shouldn't be possible.
He also said that just because doping can't be proven,
it doesn't mean it's not happening. Many cyclists in the
Tour de France this year also performed better than Peak Armstrong,
so of course people are going to wonder about the
return of widespread doping, right.
Speaker 1 (13:05):
But I assume when The Atlantic runs with the headline
science is winning the Tour de France, they're not talking
about the science of blood doping.
Speaker 2 (13:12):
That's correct. The Atlantic makes the case that the entire
sport of cycling has been transformed by science and technology
in the last twenty years, so much so that non
doping athletes can far exceed Lance Armstrong's performance without needing
to cheat.
Speaker 1 (13:26):
I mean, that is just fascinating.
Speaker 2 (13:27):
How so part of it is that they have computers
on their like literal computers on their bikes now, so
they can much more accurately track their heart rate, speed,
something that I never thought about, pedaling wattage peddling wood yes,
which Lance had a very good pedal wat I guess
you could say, and other metrics in real time. They
collect a ton more data than they used to and
(13:50):
that allows them to optimize their training and even racing
better than ever before.
Speaker 1 (13:55):
You know, it's interesting. We had Nick Thompson, the CEO
of the Atlantic and the formatagure of Wired, on the
show not too long ago, who's a very avid runner,
and he was talking about exactly this, like he has
so much data that he's doing these you know, runs
totally optimized by technology.
Speaker 2 (14:10):
It's better running through data optimization essentially. There are also,
to your point, simulation technologies that help cyclists practice parts
of the course before writing these roots in real life.
Not to mention, weather forecasting has come a long way
and can predict things like wind speed ahead of time,
which allows teams to break out the right aerodynamic gear
(14:30):
on any given day. You've seen those helmets, Like everything
from cycling jerseys, shoes, socks, and the bikes themselves are
all heavily tested in wind tunnels and tweak to achieve
as little drag as possible. I just love sort of
imagining these cyclists in these simulations looking a little bit crazy.
Speaker 1 (14:48):
But what's driven the embrace of technology in cycling was that,
like a moneyball moment where the nerves took over the
sport and displaced the jobs.
Speaker 2 (14:57):
What happened here, actually, yes, in the early in two
thousand's your home, British Olympic Cycling team develop this philosophy
of what they call marginal gains, which is, like Moneyball,
you incrementally improve everything and hope that results in larger
improvements when it's all combined. So because of that approach,
the Brits actually won a bunch of Tor Defran's titles
(15:17):
in the twenty tens as the doping lance era faded
out and then the rest of the sport kind of
followed the lead of the British Olympic Cycling team The
most interesting part of the piece and the reason that
I wanted to share it comes from this line. Today's
generation of rising stars are digital natives, for whom ignoring
the data and the appse is unthinkable. You can't win
(15:38):
without them, which is just really interesting. In terms of sport,
I think, or as you would say sport, I call
it sports, but this idea that you have to be
good at the computer essentially to be good at sports.
Speaker 1 (15:49):
Yeah, and to me, this is a very US story
because it's yes about technology bustles, so about how technology
is integrated into culture and human psychology.
Speaker 2 (15:57):
You're not allowed to opt out, No, have to be
good at this. And you know, the point of the
piece is really that a bunch of technological developments existed,
but it took a new generation of writers to embrace them, or,
as The Atlantic puts it, because writers tend to be conservative,
even superstitious in their loyalty to tried routines. Shifting the
culture took some time, and it's it's working so well
(16:19):
that people look like writers look like they're cheating.
Speaker 1 (16:21):
Which maybe they are.
Speaker 2 (16:22):
I know that we don't. We don't really know.
Speaker 1 (16:24):
The history will be their judge. So what's the takeaway
for you here? Really? I mean, obviously I remember back
in the Lance Armstrong days, people were like he has
a special way of peddling, and like that's what it's
all about. Do you think we will look back on
this and say, oh my god, that the real science
and tech innovation was around and you kind of doping
or do you buy that this kind of like the
Atlantic article suggests, collection of tech innovations when they coalesce
(16:48):
and a widely adopted can transform human performance.
Speaker 2 (16:51):
I mean, I think this is happening in every sector,
not just in sport. Which is this idea that if
you don't know how to use the newest technology, you
will all behind these days if you are going to
perform in any realm, if you are not consistently optimizing yourself,
you're kind of left in the dust. And I think
the same is true and quite literally true in cycling.
Speaker 1 (17:20):
After the break, erasing history at Google, digitizing plant pollen,
and radioactive rhinos plus, a listener calls in to our
Chat and Me segment Stay with us, Welcome back. We've
(17:48):
got a few more headlines for you this week.
Speaker 2 (17:50):
And then a Chat and Me segment submitted by chat
gbt itself.
Speaker 1 (17:54):
Before we launch into the headlines, I'd love to remind
listeners that we really do want to feature you in
the chat segment. So if you found yourself turning to Chachibt,
grock Claw, Gemini, or any other chatbot for help with
an unusual task or to answer life's deepest questions, please
send us a one to two minute voice note to
tech Stuff podcast at gmail dot com. But first, Kara,
(18:17):
we talk from time to time about how Google's search
business is under threat from chatbots. Chatchibt recently reached a
billion daily searches. Do you know how many searches there
are daily on Google?
Speaker 2 (18:30):
Three billion?
Speaker 1 (18:31):
Well, fourteen billion, geez. At least that's according to a
researcher called rand Fishkin, whose work I've found on Google
Search and amidst all the excitement about AI, Google Search
does kind of remain the world's source of truth, which means,
of course there are powerful incentives to mess with it.
Speaker 2 (18:49):
Is there a specific scandal a bruin around search?
Speaker 1 (18:53):
Yes, there is, a couple of journalists recently discovered a
way to hack Google Search that could have some serious
implications for censorship. Say more so, it turns out it's
relatively easy to basically trick Google into suppressing information by
removing certain pages from search.
Speaker 2 (19:12):
Which doesn't sound good. What does that mean?
Speaker 1 (19:14):
Well, it's good if you run a crisis pr film,
I guess. But the crazy part to me about this
story is it's being reported only because a journalist discovered
it by accident.
Speaker 2 (19:25):
Is this a happy accident or a sad accident? How
did this happen?
Speaker 1 (19:28):
Well, it's a happy accident. I suppose that the information
has been revealed. The information itself is kind of sad
and disturbing, which I'll get to. But how did he
find it? I mean he spends time, I guess, like
I do, googling himself. No, no, I'm kidding. The journalist is
called Jack Paulson. He is an investiative journalist who covers tech,
and he was looking for an old article on Google
(19:48):
when it turned out that wasn't there. Even when he
searched the exact headline of his story in quotation marks,
he discovered that it wasn't there, and in turn, he
discovered this somewhat unknown SEO trick whereby anybody who wants
to can essentially delist a page from Google Search. How
it's a little complicated, but bear with me.
Speaker 2 (20:11):
It may.
Speaker 1 (20:11):
Google has this tool called refresh outdated Content. It essentially
requests form to have Google updated search results for pages
or images that no longer exist. But anybody can do this, yeah,
basically to delist broken links. It turns out though, that
if you submit a bunch of links to a website
which has the same URL but with different letters capitalized,
(20:34):
you can direct Google to what it thinks are broken links,
because if a capitalization is wrong, you don't get to
the page you want to go to, even though it
has exactly the same letters. And if you route Google
to multiple pages that don't exist, Google can essentially be
tricked into delisting all versions of the page, including the
actual functioning one with the right letters capitalized. This is
(20:56):
basically like data poisoning Google, and it means you can't
find the functioning page on Search, which therefore renders it
more or less undiscoverable.
Speaker 2 (21:05):
And so someone had been doing this with Paulson's articles?
Is that how he figured this out?
Speaker 1 (21:09):
Yes, exactly so, the articles he was trying to find
from twenty twenty three, and as I mentioned, it's a
little bit dark, but they related to this domestic violence
charge that was pressed against a tech ceo called Delwyn
Maurice Blackman. Black Men had tried all sorts of ways
to keep the story off the internet, lawsuits, DMCA requests,
but nothing worked. So when Pulsen realized that across his
(21:33):
archive there were exactly two articles that had been d
index that were missing from Google Search, and both of
them were about black men, alarm bells started ringing.
Speaker 2 (21:43):
Seems fairly targeted.
Speaker 1 (21:45):
Yeah, so he reached out to this friend of his
who'd written an article about Blackman's attempts to suppress Pulson's reporting.
Guess what, that page, that story from the Freedom of
Press Foundation, had also been d index. So it was
clear something was afoot and it was actually the Freedom
the Press Foundation who uncovered the refreshed, outdated content trick.
Speaker 2 (22:05):
So is Google speaking out about this at all?
Speaker 1 (22:08):
Yeah, and they say they fixed the bug, but they
haven't given any real details or said how many pages
were affected. The number they offered was quote a tiny fraction.
Even if that's true, Paulson is right to point out
it's a major problem. He told four a for Media quote,
if your article doesn't appear on Google search results. In
many ways, it just doesn't exist, and he said it's
(22:28):
dangerous that a bug like this could be so easily
exploited by people in power.
Speaker 2 (22:32):
I do have some good, less daunting tech news to share,
and it's about AI and I don't know if you're
allergic to this, but pollen.
Speaker 1 (22:41):
I am. I mean it's August now, so I'm good. But
March April May, I'm suffering. Telling me as an aiqre.
Speaker 2 (22:47):
I don't know if we've come that far yet, but
the Smithsonian Tropical Research Institute is digitizing over eighteen thousand
images of pollen and then using them to train a
machine learning model to identify them.
Speaker 1 (22:59):
And help us alergy sufferers.
Speaker 2 (23:01):
Well, funnily enough, I think it could. Eventually. This database
is going to save pollen experts. And there are pollen
experts literally hundreds of hours of work hunched over their microscopes,
which might free up enough time for them to cure
your allergies. It's also going to make a lot of
new types of pollen analysis possible, and you and I
(23:24):
will be able to access this database.
Speaker 1 (23:26):
Wow, we'll be we have we have years of entertainment
ahead of us. But I guess I would have thought
by now that these scientists we're looking at these pollen
grains on a computer rather than hunched over an Industrial
Revolution style microscope.
Speaker 2 (23:38):
I mean, I imagine they can, but researchers still have to
identify pollen grains literally one at a time. So the
forty million photos being uploaded by the Smithsonian are going
to be a huge help either way.
Speaker 1 (23:51):
What are they saying this will help achieve?
Speaker 2 (23:53):
So the Smithsonian houses one of the largest pollen collections
in the world, and pollen is kind of like a
fingerprint of a specific time and place. Some grains can
last hundreds of millions of years, and each species pollen
is completely unique. So a database like this can do
everything from helping understand exactly which kind of pollen is
causing allergies to identifying where clothing found at a crime
(24:17):
scene came from. And it could even help scientists understand
how prehistoric plants adapted to climate change.
Speaker 1 (24:23):
Huh. While we're on the topic of tech and nature,
we can't pass this week by without talking radioactive rhinos.
Speaker 2 (24:32):
You're talking about rhinoceros, Yes, I am, so please tell
me more about this.
Speaker 1 (24:37):
So radioactive rhinos are actually more than a catchy name
for a band. Scientists are using radioactivity to fight illegal
rhino poaching. The best part is they're calling it the
risotope Project.
Speaker 2 (24:51):
I love that play on words, but how does the
risotope project work?
Speaker 1 (24:56):
So basically, researchers in South Africa have launched the initiative
to pla aren't low level radiation in a rhinos horn?
And then these isotopes can be detected by the equipment
at customs whenever you enter a new country, in the airport,
in a port, wherever it may be, and this in
turn will make it a lot easier to find and
confiscate rhino horns and hopefully ultimately to reduce the incentives
(25:18):
to poach them.
Speaker 2 (25:19):
I just didn't realize how many people are trying to
bring rhino horns through customs, but of course that makes sense,
and I guess my other question is wouldn't putting this
material radiation in a rhinos horn poison the rhinos?
Speaker 1 (25:33):
Well? To your first point, four hundred rhinos were poached
last year in South Africa and eight thousand across the
African continent in the last decade. This applies to both
white rhinos, which are near threatened and black rhinos, which
are critically endangered. But your second point, Yeah, obviously, the
key here is that the medicine can't be worse than
the disease, and the scientists are using very very low
(25:56):
dose radiation. They started six months ago. They've tried it
on twenty rhinos and the rhinos have been completely unharmed
by the radiation. In parallel, they tested to see if
the radiation dose was enough to trigger the detectors at
airports and stuff. It turns out the way they did
that was by three D printing a fake rhino horns,
(26:16):
putting in the same amount of radiation they put it
into the real rhino horn, and then sending a bunch
of fake smugglers through the world's ports and airports, and
it worked. Every single rhino horn got caught, including one
in a shipping container.
Speaker 2 (26:30):
So they did this with three D printed rhino horns.
Are they doing it with actual rhinos now?
Speaker 1 (26:34):
Yes. In fact, just last week they injected five more rhinos,
and they're planning to do it on a larger scale.
They're planning to partner with nonprofits to work with conservation
groups and even private rhino owners in Africa. I think
what's really interesting. We talked a bit about the platform
technology idea with cycling earlier and how these different technologies
(26:56):
came together to enhance performance. This really is only possible
because of the fact that in airports and ports around
the world there is a technology to detect people's smuggling
nuclear materials. And the head of the International Atomic Energy
Agency weighed in with what I think is the money quote.
(27:16):
By leveraging existing nuclear security infrastructure, we can help protect
one of the world's most iconic and endangered species.
Speaker 2 (27:35):
And now it's time for chatting me our segment about
how people are really using chatbots. This week, I have
a story from chat gpt itself.
Speaker 1 (27:43):
Okay, you got me, Karen, I'm very curious about this.
Speaker 2 (27:46):
So today's story is technically from a listener named Caitlin.
Hi Caitlin, but she actually asked chat gpt to write
her a detailed summary of how she uses chat gpt,
and chat gpt obliged.
Speaker 1 (27:59):
I love this. You came up with the chat and
me contain a Cara and how fun within just a
few weeks to see our listeners stretched to streached to
its limits by putting chat itself on the assignment.
Speaker 2 (28:11):
It's very very cool. It's something I would never think about.
So here she is reading what chat wrote.
Speaker 3 (28:18):
Caitlin uses chat GPT as a multi purpose thinking partner,
emotional sounding board, and creative assistant. She often turns to
it to process complex feelings, especially within her marriage and
family dynamics. She values responses that reflect honesty, nuance, and
emotional clarity, especially when exploring her relationship with her husband,
(28:38):
where she prefers a lens that aligns with his logic
and language rather than clinical terms.
Speaker 1 (28:43):
You know, I love this and thank you Caitlin for submitting.
I've heard a few friends recently talk to me about
how chat is entering the marriage and giving partner's advice
on how to interpret each other's behavior and how to
interact with each other. So very very curious to whoever goes.
So many people are making sense of their relationship with
chat GPT. And also usually it's in your corner when
(29:06):
you're the user, which is always nice.
Speaker 2 (29:09):
And CHATGPT was really thorough. It keeps track of all
of her conversation topics.
Speaker 3 (29:14):
She uses chat GPT to track and reflect on patterns
in her mood and mental cycle, explore ideas around purpose
parenting and identity, and talk through moments of loneliness, frustration,
or self doubt. She also asks for help in reframing conversations,
drafting texts, or understanding her own thoughts more compassionately.
Speaker 2 (29:33):
So chat knows what chat does, which is really interesting.
Speaker 1 (29:37):
Read Hoffmann. I she wrote a piece of The New
York Times a few months ago wasted. An interesting experiment
is to ask chat to draw a picture of you
and in my case, who was an old lady with
a cat. But I don't know it. I should do
it again to see what it is now. But yeah,
I mean, this is like it reminds me of Harry
Potter and The Pensive, which was this kind of bucket
where Dumbledore could point his wand and see other people's memories.
(29:59):
Remember this, I have no.
Speaker 2 (30:00):
Idea what you're talking about, but it does feel like
chat or Caitlyn's chat GBT is looking into her soul.
Things get a little messy for Caitlin, and there's an
error in the next section.
Speaker 3 (30:11):
Beyond emotional support, Caitlin uses chat GPT creatively for mural planning,
brainstorming business names, home improvement designs like pergolas, or organizing
craft supplies. She draws on it for lifestyle help too,
like shopping for swimsuits, reducing caffeine, or managing family logistics.
Speaker 2 (30:30):
So as guess which part of this description is not accurate.
Speaker 1 (30:33):
I mean it feels one of those kind of IQ
slash logic tests. My first guess would be that reducing
caffeine intake is unlike the other activities in this list.
Speaker 2 (30:44):
So no, it's not that wrong, and it's not the
Pergola designs, which I guess CHATGBT helped with. CHATGBT got
a little cocky because Caitlin says she never asked it
for ideas for her murals. She only had uploaded pictures
of her previous work to get feedback. So chat had
nothing to do with the mural design.
Speaker 1 (31:01):
Nothing to do with coming up with the ideas, only
giving her feedback. That's interesting, So chat back in your place.
Speaker 2 (31:07):
Well, for Caitlin, it's not chat, it's Gary. Caitlin also
told us via email that she named her chatbot Gary
and sees him as a non threatening male character to
talk to sometimes. We all need a non threatening male
character to talk to sometimes, and Gary had the following
to say about the way Caitlyn uses chat GBT.
Speaker 3 (31:25):
Overall, underneath it all is a thoughtful desire to grow,
express and stay grounded, often in the face of competing
responsibilities and emotional weight.
Speaker 1 (31:34):
Thank you, Caitlin. This is a portrait of a life
and a person harnessing technology and digital tools to help
them live a better life. And it's also a fascinating
insight into just how much our chatbots know about us,
even if they sometimes fabricate at the margins, Kaitlin, those
murals are yours.
Speaker 2 (31:54):
And yours alone. Caitlin, thank you so much for sending
this voice memo text of listeners. Love having your voices
on our show, and we'd love to hear more peculiar
or useful ways you're using chatchypt Grock, Claude, Gemini, or
any other chatbot. Remember you can send us a one
to two minute voice note to tech Stuff podcast at
(32:14):
gmail dot com. That's it for this week for Tech Stuff.
Speaker 1 (32:33):
I'm Kara Price and I'm mos Valoshin. This episode was
produced by Eliza Dennis and Tyler Hill. It was executive
produced by me, Kara Price and Kate Osborne for Kaleidoscope
and Katrina novelfa Ihart Podcasts The engineer is Katherine Cook
and Jackinsley mixed this episode. Kyle Murdoch rodel theme song.
Speaker 2 (32:51):
Join us next Wednesday for tex Stuff the Story, when
we will unpack how Google Search got a hold of
your CHATCHYPT conversations.
Speaker 1 (32:59):
Please rate, review, and reach out to us at tech
Stuff Podcast at gmail dot com. We love hearing from you.
TechStuff News
Hosts And Creators
Oz Woloshyn
Karah Preiss
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Special Summer Offer: Exclusively on Apple Podcasts, try our Dateline Premium subscription completely free for one month! With Dateline Premium, you get every episode ad-free plus exclusive bonus content.
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!