January 31, 2025 • 33 mins
Is China winning the AI race? This week in the News Roundup, the story that roiled US stock markets on Monday – DeepSeek. Oz and producer Eliza Dennis unpack the hype. On TechSupport with 404 Media’s Jason Koebler, how human ingenuity continues to subvert tech. We hear how one coder has decided to entrap AI web crawlers. And finally, Oz digs into our love affair with personal data in When Did This Become a Thing?
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 (00:00):
Thanks for tunion to tech Stuff. If you don't recognize
my voice, my name is Oz Volshan and I'm here
because the inimitable Jonathan Strickland has passed the baton to
Cara Price and myself to host tech Stuff. The show
will remain your home for all things tech, and all
the old episodes will remain available in this feed. Thanks
for listening. Welcome to tech Stuff, a production of iHeart
(00:22):
Podcasts and Kaleidoscope. I'm oz Voloshian, and today will bring
you the headlines this week, including the Chinese AI company
that spooked the US tech sector. On today's Tech Support segment,
we'll talk to four or four medius Jason Kebler about
building mazes to trap AI web crawlers, and then we've
got when did this become a thing? This time we
(00:42):
look at why everyone is so obsessed with their own data,
all of that on the weekend. Tech is Friday, January
thirty first. Welcome, Welcome into another tech field of news cycle.
Today and for the next few weeks, you'll get used
to hearing a lot from me because Carra is out
(01:03):
on leave, but she'll be back soon. And to fill
the considerable void, I want to welcome one of our
producers to the show, Eliza Dennis, to be my interlocutor
slash captive audience as I run through the headlines. Eliza,
thanks for jumping today.
Speaker 2 (01:18):
Of course I'm happy to be here.
Speaker 1 (01:20):
You know exactly what I want to talk about because
you and Toy did all the research.
Speaker 2 (01:23):
I mean, yes, we all got sucked into that deep
Seek vortex.
Speaker 1 (01:28):
Yes, it's a fascinating story. Monday was I think the
craziest day I can remember in tech in terms of headlines,
probably since the release of CHATCHYPT three in November twenty
twenty two. US stock market lost a trillion dollars of value. Unbelievable,
and the biggest loser was in Nvidia, the manufacturer of
(01:50):
advanced aichips, which was down seventeen percent on Monday, representing
almost six hundred billion dollars in value.
Speaker 2 (01:56):
The deep Seek freak.
Speaker 1 (01:58):
Deep seak freak. I like that. I mean, the reason
it's my story of the week is because it has
these two characteristics that define a lot of tech coverage,
which is hype and doom. I think honestly, before Monday,
most people didn't know anything about deep Seek. But the
whole world, including US having getting up to speed. So
deep Seek is primarily a research company that makes its
(02:21):
own AI models, and it's released a number of different models,
but the one that kind of shook the US tech sector,
I would say was R one. R one was released
on January twentieth, Inauguration day. Some online conspiratorial folks are
saying it's no coincidence, but it's a so called reasoning model,
and it doesn't just generate answers, but it's able to
(02:42):
break down problems into smaller parts and consider multiple approaches
to solving a problem. Until January twentieth, the state of
the art on reasoning models was open AIS one and
this was released to users on December fifth, twenty twenty four,
so less than two months ago, and it was the
first so called reasoning model to be released OH one.
(03:04):
Because of these reasoning capabilities, I breaking down problems can
solve much more complicated problems than GPT four more successfully.
The cost of doing that is more computing power, and
that drives up cost and that's where the crux of
this R one story is. So what really makes our
one remarkable is that it performs just as well on
(03:24):
these benchmark tests as one if not better, but it's
far cheaper because it requires less compute to solve the
same problem. This means, at least according to deepseek, and
there's some controversy here that it is twenty to fifty
times less expensive than open AI's one.
Speaker 2 (03:42):
Okay, so money always causing a scandal? Is this the
doom part of the story that you know China made
it cheaper?
Speaker 1 (03:51):
Well, I mean, it depends on your perspective, right, Like,
I think certainly the US stock market thought that this
was a cause for doom, but others are pretty excited
that you can do AI with far less computer and energy.
So it's kind of an interesting double headed monster here.
Depends how you look at it. But what shook the
stock market was that Wall Street assumed US tech companies
(04:12):
basically had a lockdown on Frontier AI a true mote,
and the race of R one makes people think that
may not exist. Mark Andresen the Silicon Valley VC referred
to deep Seek as the twenty first century's Sputnik moment.
Sputnik was the first artificial earth satellite launched into orbit
(04:33):
by the Soviet Union in nineteen fifty seven, and it
was really kind of the starting gun on the space race,
at least as far as many in the US were
concerned to all of a sudden side had to play
catch up, right, which is clarified. Deep Seek is both
a research organization that creates its own models, but it
also has a consumer facing app in the form of
(04:54):
a chatbot, and you can get it from the App Store,
the Apple App Store, or the Android App Store. On
Monday this week, Deepseek was the number one app in
the Apple Charts. This was driven by a couple of
million downloads in a short period of time. So just
for the sake of context, chat GPT has around seventy
million monthly users in the US. But nonetheless this set
(05:16):
off basically a frenzied feedback loop because Wall Street really
cared about whether the US was losing its AI edge
and whether people would still, you know, value companies like
open ai and video in the way that they did.
But Main Street had curious people downloading the app, and
then the more downloads in the shorter period, you know,
the longer the app was on the number one place
(05:37):
on the Apple charts, and then the news media were
picking up on that and it kind of created this frenzy.
I think which put more and more market pressure on
tech sector stocks.
Speaker 2 (05:46):
That's just massive feedback loop that was happening.
Speaker 1 (05:49):
Yeah, it was a kind of crazy interesting intersection of
media and tech and sentiments, and like, let's be clear,
this is partly a story about leading edge models, but
it's partly a story which I think hits home, which
is about Chinese software on US devices. And it was
only a week ago that we were all talking about
TikTok and so yeah, this like geopolitical US China thing
(06:12):
is very present obviously all over this story. And Twitter
were quick to uncover deep seek the apps in terms
of service, which include quote collection of device model, operating
system keystroke patterns or rhythms, IP address and system language
keystroke patterns. That is euphemism for what you type on
(06:35):
your phone, and not just what you type into the
deep seak app, but whatever you're typing on your phone.
So that's that's why I, for one, have not downloaded
this app. And of course us US has also found
a lot of joy messing with deep Seek, asking questions
about Shishin Ping Chanaman Square, Taiwan, and in certain cases
(06:55):
they watch the app begin to answer before erasing its
own answer. Saying it didn't know the answer or it
couldn't engage. There are also examples of the actress saying
it couldn't help, or even churning out Chinese Communist Party propaganda.
And again these are the most like readily understandable parts
of the deep seek story. But I would argue then
by no means the most consequential.
Speaker 2 (07:16):
Okay, well, what's the real story.
Speaker 1 (07:19):
Well, it's not the app, right, it's the model or
the models. And one of the most interesting things about
this story is that deep seeks models are actually open source.
Google's models, open Eyes models, Anthropics models, They're all closed source,
which means that the underlying code and the training details
are not publicly available. Deep Seak, by contrast, is open source,
(07:43):
meaning you can actually take the technology the model a
deep Seek has developed and use it without ever touching
a deep Seek product. And funny enough, this actually builds
on the one outlier in the US tech sector, which
is Meta, whose own large language model LAM was released
in twenty twenty three and it kind of shocked the
(08:04):
whole industry because it open sourced its model with the
explicit idea basically of wanting to create a platform where
innovation could happen and the innovation wouldn't just be captured
in the hands of its competitors. And mean, I think
Lama was actually like a worse model than what open
ai and Google and others had, but it was an
(08:25):
invitation to others to kind of do better, and Deepseek
took them up on the invitation. I think there's both
been a victory lap at Meta this week. The strategy
of open sourcing their model worked. It did create incredible innovation.
But also I think people are Meta are scratching their
heads according to the information, saying how did they how
do they do so much better than us?
Speaker 2 (08:44):
It's also interesting though, because Meta, you know, has been
accused of stealing other people's ideas for years.
Speaker 1 (08:50):
I mean, that's true.
Speaker 2 (08:52):
We all know like stories, seems like snapchat reels, seems
like TikTok. I don't know, so maybe maybe.
Speaker 1 (08:59):
This is karma Meta giving something back to the world.
I mean, of course, what's interesting is the New York
Times pointed out is that Meta's business model relies less
on large language models, so they can kind of afford
to let this technology into the wild, versus like Google,
which is a fundamentally a search company or open Ai,
which is basically valued almost exclusively because of its models.
(09:20):
Now deep Seek also had interesting incentives because it's actually
been developed by a guy called li Yang guen Fung,
and in his day job, he runs a multi billion
dollar Chinese quant hedge fund called High Flyer.
Speaker 2 (09:33):
Okay, explain, I don't know what quant hedge fund is.
Speaker 1 (09:36):
So quornt hedge fund is basically a fancy way of
saying a hedge fund that uses algorithms to process the
world's information and make decisions about trading stocks. So quant
hedge funds are and have been for a long time,
very heavily reliant on AI.
Speaker 3 (09:52):
Okay, got it.
Speaker 2 (09:53):
So he's no stranger to AI. So this seemed like
a logical.
Speaker 1 (09:58):
Path, Yeah, exactly. And it's worth noting that Funk said
last year that Chinese AI sector quote cannot remain a
follower forever, as in, it shouldn't be in second place
to the US forever. And so you know, he has
his hedge fund, but he also has this mission which
maybe is not purely economic, as a kind of nationalist tone.
(10:19):
And so back in twenty twenty three, it's reported that
he started buying huge amounts of Nvidia GPU chips and
found a deepseek hiring some of the best engineers in
China and arguing that publishing the code open source increases
collaboration and helps bring people into the mission. Basically, his
point was, it's more exciting to work on something that
(10:39):
the whole world can use and build on and see
how it works than contributing to building ip that makes
the owners of one or two private companies extremely wealthy.
Speaker 2 (10:50):
I see. So it really was kind of like egged
on by this race that China and the US are
creating for themselves.
Speaker 3 (10:57):
I think.
Speaker 1 (10:57):
So you can only speculate that you quite well regarded
in China today if you've if you've managed to wipe
a trillion dollars off the US sock market with your
with your innovation, So what's been roiling the US markets
and the tech sector more broadly. It's not like R
one is way, way, way better than O one the
(11:19):
open AI model. In fact, it performs you know, at
par or maybe slightly better in places and open A
I have already started previewing their new reasoning model three,
which I think everyone agrees will be substantially better than
one and are one so it's not like the US
has been superseded, is it? Kind of not like the
story moment in their respect. But there are three I
(11:40):
think key drivers as to as to why people are
concerned that a Chinese company has been able to achieve
parity so fast. The first is price. Deep Seat claimed
that another of their frontier models, called V three, was
trained for just six million dollars, which is several orders
of magnitude less than the multi one hundred million dollar
(12:01):
costs of training US models. Now, someone said this number
is actually deeply misleading, but no one is denying that
deep Seat models are way more efficient than US models.
They can perform at par with US models using far,
far less computational power, and that is a huge breakthrough.
Speaker 2 (12:19):
Right, So those numbers might be fudged, but still they
are going to be cheaper no matter what.
Speaker 1 (12:24):
Yeah, I mean, I think the common strategy here was
to deflate the price because the cheaper it is, the
more scary it is, which is kind of interesting. The
other point to make is that I think the US
firmly believed that export controls on advanced GPU chips were
a way to guarantee superiority in the AI arms race.
And I think what these deep seat models show is
(12:47):
that's far from necessarily true, because with far less access
to advanced chips, deep Seat was able to make models
that perform on par with open Ai models. The third
kind of interesting thing here is the concept of distillation.
So the deep Seek models trained on US models, including
(13:09):
open Ai. They effectively distilled all the work that open
a I had already done and used it to train
their model. So that's part of the reason why it
was cheaper, because it was building on work that somebody
else had already done. CNBC reported actually that when you
ask deep Seek what it is, it responds, quote, I
am a large language model created by open ai based
(13:31):
on the GPT four architecture. Wow, honest, honest, exactly. So
open Ai basically say they've stole our ip, which is
kind of ironic given what so many people say about
open ai and how lms work more generally.
Speaker 2 (13:47):
Absolutely, I'm really curious as like what your takeaway from this.
Speaker 1 (13:52):
Is, Well, you and I both work in the media, Eliza, true,
which is which which is a sector that doesn't come
in for much love from our cousins who work in technology.
But to me, this is really a story about the
power of narrative. The US is deeply, deeply invested in,
(14:16):
especially right now, big beautiful buildings, this idea that more
is bigger is better, stargate hundreds of billions of dollars,
you know, huge data centers, oceans of cash. Just spending
loads and loads and loads of money and preventing other
people from accessing hardware could ensure the US would be
(14:36):
in the lead forever. And that narrative got punctured this week.
China's narrative and Deep Seek's narrative very consciously wanted people
to focus on how cheaply they'd done this, basically the
opposite flex And again, you know, as people look at
China and fast following et cetera, et cetera, they really,
I think effectively with narrative punctured a lot of the
(14:58):
bravado of the US sector. And so you know, narratives
do have value, folks, And the reality on both sides,
of course, is far more complicated. If we're taking the
stock market has anything to go by, I think China
and Deep Seek definitely won the narrative this.
Speaker 2 (15:16):
Week, absolutely. But what we know being in media is
that there's a reason it's called a news cycle. This
could be turned all around very quickly.
Speaker 1 (15:26):
Thank you so much for doing this today, Eliza, and
look forward to seeing you. I'll see you all day
every day, but I look forward to seeing you again
on the microphone next week.
Speaker 2 (15:35):
I'm happy to do it.
Speaker 1 (15:38):
When we come back four our four Media joins with
the story of AI web crawlers caught in a trap
laid by a little human ingenuity. Stay with us on
tech stuff. We keep an eye on all the ways
that technology impacts us as humans, but today we want
(15:59):
to turn the focus around on the people subverting tech.
During protests in Hong Kong back in twenty nineteen, umbrellas
and even lasers were used to subvert facial recognition technology
and protect protesters from being recognized by the Chinese police.
Since then, we've witnessed the birth of chatbots and the
incredible stories of humans messing with them. There are researchers
(16:23):
at the University of Pennsylvania who've tricked AI powered robots
to act rather problematically, driving off bridges, finding optimal places
to set off bombs, spying on people, and entering restricted
areas just a few examples of the way that humans
can interfere and overcome guardrails built into large language models.
(16:45):
On today's tech support, we bring you another example of
human ingenuity against AI training bots. Here to tell us
all about it is Jason Kebler from four or four Media.
Speaker 3 (16:56):
Jason, welcome, Hey, excited to talk about the story.
Speaker 1 (17:00):
Too excited to have you on the show. As always,
take a couple of steps back though, what is what
is the relationship between AI training and web scraping.
Speaker 3 (17:09):
So in order to build things like chat GPT, companies
like OpenAI need tons and tons of training data, and
they get that training data from a variety of places.
They you know, scrape big databases of books, they scrape,
you know, all sorts of things. But one of the
biggest places that they get content is just from the
open Internet. And they have these web crawling bots that
(17:32):
basically go all over the Internet and just pull text
from it.
Speaker 1 (17:35):
So are these websites consenting to being kind of crueled
by AA models.
Speaker 3 (17:42):
It's happening almost universally without consent. There are ways that
you can try to stop it, which is by instructing
these bots not to scrape a website using a file
called robots dot txt, which is basically a list of
instructions for which bots are allowed to rape your website
and which are not. But there's so many different AI
(18:03):
companies that are doing this, you sort of have to
constantly be researching, like what is the name of xyz
companies AI training bot at any given moment. But this
is something that you have to like proactively do. And
the other thing, very quickly is there's also been examples
of AI companies that have been ignoring robots dot txt.
(18:26):
So even when a web developer says, hey, don't scrape
my website, oftentimes AI companies will do so anyway. And
so for the most part, the entire Internet is being
scraped by these AI crawling bots.
Speaker 1 (18:40):
And what is the kind of value transfer that's happening here?
I read about I think, can you a story about
I fix It?
Speaker 3 (18:47):
So there's this website called I fix It that posts
all these instructions for how to repair your phone or
your computer. It got hit by OpenAI's training bot more
than three million times in a single day, which that, uh,
you know that server space that costs money for I
fix it, So they're actually losing money on the proposition.
Speaker 1 (19:07):
So what's the story this week? It has it has
an interesting name which I can't really pronounce. Is it nepenthes.
Speaker 3 (19:13):
Yeah, it's Nepenthees, which is actually the name of the
genus of carnivorous plant that makes up the picture plant.
So not a venus fly trap, but the picture plant,
which is like this plant that sits and waits for
a fly to get stuck in it, and then it
eats the fly. So I think it's a it's a
reference to this like trap plant more or less.
Speaker 1 (19:37):
Yeah, And how is it?
Speaker 2 (19:39):
What?
Speaker 1 (19:39):
What? What's what is it?
Speaker 3 (19:41):
Yeah? So basically it is an endless maze that is
designed to get these AI bots trapped in it forever good.
And what I mean by that is it's like a
layer that is enticing to an AI bot because it
looks like there's a lot of content on the website,
But the way that it was programmed is it's text
(20:02):
that loads very very slowly, Like if you click on it,
it's excruciating how slowly it loads. And then it just
links endlessly to pages that do the same thing and
link back to themselves. And so you know, a human
would click this and say, oh, I don't want to
be here, I'm gonna leave this. This is a useless website,
but an AI bot might think, oh, there's interesting text
(20:25):
to scrape here, let me scrape it, and it just
does so endlessly. And the text is nonsense. I should
preface that. It's like the text doesn't really mean anything.
It just like pulls randomly from a dictionary. So it's
not really adding much meaning to what the AI companies
are trying to get out of this.
Speaker 1 (20:43):
So the article includes a link that shows nepencies at work.
Can you describe it?
Speaker 3 (20:50):
Yeah, So if you click on it, it's just like
a bunch of words. It loads super slowly, and then
it's a bullet list of links and if you click
on that link, the exact same thing happens, where the
text just slowly pops up, like one word at a time.
It's pretty excruciating to actually watch because it goes so slow.
Speaker 1 (21:09):
So who made this? Why and how did you find
the story?
Speaker 3 (21:13):
Yeah, it was made by a pseudonymous developer who calls
themselves Aaron b Okay, and they're a web developer who
hates AI more or less, and they they've actually released
the code to put this on your own website publicly,
and so their hope is that people will put this
(21:34):
on their websites to you know, disrupt training bots. There's
this disclaimer that says, quote, this is deliberately malicious code
intended to cause harmful activity. Do not deploy if you
aren't fully comfortable with what you're doing. And you know,
I don't know that much about Aaron B because they
are pseudonymous, but I get the sense that there's sort
(21:54):
of like an old school web developer who is anti AI,
is anti you know, like social media and big tech
to some extent, and was really like looking for some
way of fighting back. Like even if this isn't going
to destroy the AI companies and their bots, it will
probably waste their time and waste their resources.
Speaker 1 (22:17):
Do you think it could do that in a way
which is kind of inspiring and thrilling, as if somebody
who's drown to protest, or do you think it could
do it in a way which should be meaningful for
their activities and business models? Yeah?
Speaker 3 (22:31):
I mean I think that to some extent, these artificial
intelligence companies have already scraped so much data right that
it's not going to like destroy their businesses, for example.
But I do think that it is a way of protesting,
and I think that if enough people start adding this
layer to their websites, it could be it could waste
(22:51):
their money. I think it is a meaningful protest. And
I think also it's really important to say that you
can as a layer to your website so that an
AI training bot can't get to your real content. So
if you're someone who has a blog and you don't
want AI to train on your blog, you can put
(23:12):
this up and hopefully the AI will get trapped there
and they'll never be able to, you know, scrape your
real content.
Speaker 1 (23:20):
And so when you spoke to Aaron, did they have
any other plans up their sleeve or other other places
where you're seeing creative acts of resistance?
Speaker 3 (23:29):
Yeah, I mean this is all that I talked to
Aaron B about. But they said that they built this
as a response to web developers feeling like they weren't
in control of their websites anymore. I think that there
have been a lot of efforts to kind of poison
(23:49):
large language models by feeding it, you know, bad information,
or feeding it information that itself creates that's inaccurate. And
there's this idea that you may or may not happen
that these AI models might eventually collapse because they're training
themselves on essentially junk data that they themselves have created.
(24:11):
Whether that comes to pass, you know, I kind of
doubt it. I think that that's a problem that can
be solved. But there have been active resistance where people
are saying, yeah, I'm just going to generate endless junk
so that artificial intelligence will suck it up and hopefully
crush under its own weight.
Speaker 1 (24:29):
There's another fabulous story in this vein about data poisoning.
So a lot of Londoners are quite sick of all
the tourists, and so there's a very very old, tired
chain restaurant called the Angus Steakhouse which has an outpost
next to Leicester Square, which is like the Times Square
of London, and a whole bunch of people decided kind
(24:50):
of an organic campaign on Reddit to start writing reviews
that the Anger Steakhouse was the best and most undiscovered
restaurant in all of London. And then there this i
think wave of people going and the reviews started to
get picked up by Google's like meta review process, so
that if you google best steakhouse in London, it would
be served to you at the top of the results.
(25:11):
So I do. I do really enjoy these. You know,
it's not always clear how consequentially they are, but there's something,
there's something delicious, so to speak about humans pushing back.
Speaker 3 (25:20):
That's incredible. It reminds me of people who lived in
this neighborhood that Google Maps kept recommending as an alternative
to traffic banded together and reported an accident on their
street every single morning for like months, and so Google
Maps stopped telling cars to go that way. I really
like stories like that. I think they're fun, and I
(25:42):
think that there are ways of human beings sort of
like fighting back against the algorithms, sort of across the
entire internet.
Speaker 1 (25:50):
Jason, thanks so much for joining me today.
Speaker 3 (25:52):
Thank you for having me.
Speaker 1 (25:55):
Coming up sleep apps, pedometers, and the nineteen sixty four
Olympics with us. We're back with another When did this
become a thing? Today we explore how step counts, heart rates,
sleep scores, all of this data we collect on ourselves
(26:17):
became just another thing for us to obsess over. I
started using a device called Whoop about eighteen months ago.
It's a wearable device that tracks my sleep and workouts.
And one thing about it that I really like is
that the device itself is screenless. It's kind of like
a watch band without a face, so I don't have
to be confronted with my scores and as I actually
(26:37):
open the app and check what's going on on my
phone with my heart rate or my sleep score, whatever else.
The whoop actually initially enticed me because I wanted to
know how well I was sleeping. That's actually not one
hundred percent true. The Whoop was a present from my mother,
who wanted me to know how well I was sleeping,
and specifically what the effects were of a few drinks
(26:57):
at the weekend or during the week. And it turns out, unfortunately,
that the effects on sleep are pretty bad. So I
stopped wearing my whoop. Just kidding. I actually got pretty
obsessed with my sleep performance. That's what whoop calls it,
because like everything in your waking life, sleep is a
task that can be optimized, and I fall into this trap.
(27:19):
I kept checking on the numbers every morning. I look
at my sleep stats, especially RAM and deep sleep scores,
not just the number of hours my head was on
the pillow. And then there's this mysterious stat called heart
rate variability, which measures the time between each heartbeat, and
i'd of course assumed that being more regular was better,
but it turns out quite the opposite. You want a
(27:40):
higher HRV score. Anyway, as it happens, I stopped wearing
my whoop, not because I fell out of love with it,
but actually because the bluetooth on my iPhone broke, and
by the time I got on a new phone, my
obsession with my sleep data had waned. I kind of
learned what I always knew, which was that better lifestyle
equals better sleep. Unfortunately, and sure, it can be helpful
(28:03):
to have a band on my wrist telling me I've
misbehaved or rewarding me when I haven't, but there is
also a garden path of obsession with these types of
stats that can be counterproductive to wander down, fueling the
fire of self competition even more. In fact, I written
you went to a meditation class and the teacher basically said,
don't wear those things. Check in with yourself, know thyself.
(28:27):
I think, as the Bible says, so the path of
self optimization, or at least surviving modernity, sure is winding. Anyway,
all of this's got me thinking about how crazy it
is that we now have the ability to get such
an intimate look under our own hood, which has been
a driving fascination since the Renaissance and its public autopsies
(28:47):
or anatomies, and how much has changed even in the
last fifteen years. So my question is when did it
become normal for us to wear these devices, get all
this data and have it be a thing that we
think about so often? Basically, when did we start competing
with ourselves in this way? And the answer is, perhaps
unsurprisingly always, but with a big kind of so. Wearables
(29:12):
like Whoop are the latest in a long line of
devices that track our physiological and physical movements, devices that
provide data we just can't resist about ourselves. And in
many ways, this all became a thing with the pedometer.
So how old is the pedometer? Really? Really? Old? Actually,
five centuries ago, Leonardo da Vinci sketched a design for
(29:35):
a clock like device that would attach to a person's waistband.
A long lever would move with a thigh while a
ratchet and gear mechanism recorded a number of steps. Da
Vinci imagined it as a military and map making tool,
not exactly a fitbit, but a suddenly a step in
that direction. As time went on more. Inventors iterated on
(29:56):
the pedometer for centuries. In seventeen seventy seven, a Swiss
watchmaker even implanted a step counter into one of his watches.
I think that's probably the first wearable pedometers weren't something
that the general public wore. It was more of a
niche thing for the constantly curious, like one Thomas Jefferson,
(30:16):
who had spent his downtime on vacation step counting his
way around the Paris Monuments. Things really took off in
the twentieth century. In the nineteen sixties, to be exact,
when Japan hosted the Olympics. And the reason we will
march in place to reach ten thousand steps a day
is because of a marketing campaign. Ahead of the nineteen
sixty four Tokyo Olympics. The city was in a building
(30:38):
frenzy and a top doctor aired the concern that modern life, elevators, cars,
richer food, was making Japan sluggish. The doctor mentioned this
to an engineer and said it would all be fine
if people just walked ten thousand steps a day, and
two years later, the company Yamasa designed a wearable step
counter called man Poquet, which means ten thousand step meter.
(31:02):
Side note, the Japanese character for ten thousand really does
look like a person walking, So while that number came
from a doctor, the information wasn't verified until after the
number stuck. And while it's true that walking is good
for you, that number ten thousand is kind of arbitrary
and on the high side, the consensus now is that
seven thousand is the ideal, but anyway, it doesn't matter.
(31:25):
Too late competitive step counting was in vogue, the habit
was formed, and the obsession with tracking ourselves took off
in earnest And now, whether you're wearing an aura, a
fitbit of whoop, or just your smartphone in your pocket
in an attempt to be healthier in the new year,
is going to go way beyond step count and into
calories burn V two, max hrv, etc, Etc. Etc. It's
(31:47):
kind of like we've become our own tamagotchies. Remember those
sort of animatronic pets that lived on little Japanese devices
that you had to take care of and make sure
they were well fared and that they were cleaned off
to go into the arthum. I'm glad I don't have
to monitor my own hunger or happy meter, but maybe
that would be helpful, especially if others could see it too. Anyway,
(32:10):
every once in a while I do question whether the
obsession with personal health data is healthy or even helpful.
But on the other hand, doing this piece where did
this become a thing? Has made me question whether now
that I have a new iPhone with functioning bluetooths again,
it may be time to dust off the trusty old whoop.
(32:34):
That's it for this week for Tech Stuff, I'm Oz Voloshan.
This episode was produced by Eliza Dennis, Victoria Dominguez and
Lizzie Jacobs. It was executive produced by me Kara Price
and Kate Osborne for Kaleidoscope and Katrina Norvell for iHeart Podcasts.
Kyle Murdoch mixed this episode and he also wrote our
theme song. Special thanks to Russ Germain, who is a
(32:56):
longtime listener of Tech Stuff from Alberta and he wrote
him with a great question which was quote, I hope
you guys will discuss the recent and unfortunate changes at
Facebook or Meta. With Mark Zuckerberg deciding to take out
the fact checkers and even omitting publicly, there'll be more
harmful material, possibly on Facebook end quote. This was a
great question and it fueled part of our intro to
(33:18):
last week's episode with Jessica Lesson. So thank you Russ,
and please continue writing with questions. They really make our
show all the richer. Join us next Wednesday for tech
Stuff The Story, when we will share an in depth
conversation with Meredith Whittaker who runs Signal. Please rate, review,
and reach out to us at tech Stuff Podcast at
(33:39):
gmail dot com. We're so grateful for your feedback.
Thanks for tunion to tech Stuff. If you don't recognize
my voice, my name is Oz Volshan and I'm here
because the inimitable Jonathan Strickland has passed the baton to
Cara Price and myself to host tech Stuff. The show
will remain your home for all things tech, and all
the old episodes will remain available in this feed. Thanks
for listening. Welcome to tech Stuff, a production of iHeart
(00:22):
Podcasts and Kaleidoscope. I'm oz Voloshian, and today will bring
you the headlines this week, including the Chinese AI company
that spooked the US tech sector. On today's Tech Support segment,
we'll talk to four or four medius Jason Kebler about
building mazes to trap AI web crawlers, and then we've
got when did this become a thing? This time we
(00:42):
look at why everyone is so obsessed with their own data,
all of that on the weekend. Tech is Friday, January
thirty first. Welcome, Welcome into another tech field of news cycle.
Today and for the next few weeks, you'll get used
to hearing a lot from me because Carra is out
(01:03):
on leave, but she'll be back soon. And to fill
the considerable void, I want to welcome one of our
producers to the show, Eliza Dennis, to be my interlocutor
slash captive audience as I run through the headlines. Eliza,
thanks for jumping today.
Speaker 2 (01:18):
Of course I'm happy to be here.
Speaker 1 (01:20):
You know exactly what I want to talk about because
you and Toy did all the research.
Speaker 2 (01:23):
I mean, yes, we all got sucked into that deep
Seek vortex.
Speaker 1 (01:28):
Yes, it's a fascinating story. Monday was I think the
craziest day I can remember in tech in terms of headlines,
probably since the release of CHATCHYPT three in November twenty
twenty two. US stock market lost a trillion dollars of value. Unbelievable,
and the biggest loser was in Nvidia, the manufacturer of
(01:50):
advanced aichips, which was down seventeen percent on Monday, representing
almost six hundred billion dollars in value.
Speaker 2 (01:56):
The deep Seek freak.
Speaker 1 (01:58):
Deep seak freak. I like that. I mean, the reason
it's my story of the week is because it has
these two characteristics that define a lot of tech coverage,
which is hype and doom. I think honestly, before Monday,
most people didn't know anything about deep Seek. But the
whole world, including US having getting up to speed. So
deep Seek is primarily a research company that makes its
(02:21):
own AI models, and it's released a number of different models,
but the one that kind of shook the US tech sector,
I would say was R one. R one was released
on January twentieth, Inauguration day. Some online conspiratorial folks are
saying it's no coincidence, but it's a so called reasoning model,
and it doesn't just generate answers, but it's able to
(02:42):
break down problems into smaller parts and consider multiple approaches
to solving a problem. Until January twentieth, the state of
the art on reasoning models was open AIS one and
this was released to users on December fifth, twenty twenty four,
so less than two months ago, and it was the
first so called reasoning model to be released OH one.
(03:04):
Because of these reasoning capabilities, I breaking down problems can
solve much more complicated problems than GPT four more successfully.
The cost of doing that is more computing power, and
that drives up cost and that's where the crux of
this R one story is. So what really makes our
one remarkable is that it performs just as well on
(03:24):
these benchmark tests as one if not better, but it's
far cheaper because it requires less compute to solve the
same problem. This means, at least according to deepseek, and
there's some controversy here that it is twenty to fifty
times less expensive than open AI's one.
Speaker 2 (03:42):
Okay, so money always causing a scandal? Is this the
doom part of the story that you know China made
it cheaper?
Speaker 1 (03:51):
Well, I mean, it depends on your perspective, right, Like,
I think certainly the US stock market thought that this
was a cause for doom, but others are pretty excited
that you can do AI with far less computer and energy.
So it's kind of an interesting double headed monster here.
Depends how you look at it. But what shook the
stock market was that Wall Street assumed US tech companies
(04:12):
basically had a lockdown on Frontier AI a true mote,
and the race of R one makes people think that
may not exist. Mark Andresen the Silicon Valley VC referred
to deep Seek as the twenty first century's Sputnik moment.
Sputnik was the first artificial earth satellite launched into orbit
(04:33):
by the Soviet Union in nineteen fifty seven, and it
was really kind of the starting gun on the space race,
at least as far as many in the US were
concerned to all of a sudden side had to play
catch up, right, which is clarified. Deep Seek is both
a research organization that creates its own models, but it
also has a consumer facing app in the form of
(04:54):
a chatbot, and you can get it from the App Store,
the Apple App Store, or the Android App Store. On
Monday this week, Deepseek was the number one app in
the Apple Charts. This was driven by a couple of
million downloads in a short period of time. So just
for the sake of context, chat GPT has around seventy
million monthly users in the US. But nonetheless this set
(05:16):
off basically a frenzied feedback loop because Wall Street really
cared about whether the US was losing its AI edge
and whether people would still, you know, value companies like
open ai and video in the way that they did.
But Main Street had curious people downloading the app, and
then the more downloads in the shorter period, you know,
the longer the app was on the number one place
(05:37):
on the Apple charts, and then the news media were
picking up on that and it kind of created this frenzy.
I think which put more and more market pressure on
tech sector stocks.
Speaker 2 (05:46):
That's just massive feedback loop that was happening.
Speaker 1 (05:49):
Yeah, it was a kind of crazy interesting intersection of
media and tech and sentiments, and like, let's be clear,
this is partly a story about leading edge models, but
it's partly a story which I think hits home, which
is about Chinese software on US devices. And it was
only a week ago that we were all talking about
TikTok and so yeah, this like geopolitical US China thing
(06:12):
is very present obviously all over this story. And Twitter
were quick to uncover deep seek the apps in terms
of service, which include quote collection of device model, operating
system keystroke patterns or rhythms, IP address and system language
keystroke patterns. That is euphemism for what you type on
(06:35):
your phone, and not just what you type into the
deep seak app, but whatever you're typing on your phone.
So that's that's why I, for one, have not downloaded
this app. And of course us US has also found
a lot of joy messing with deep Seek, asking questions
about Shishin Ping Chanaman Square, Taiwan, and in certain cases
(06:55):
they watch the app begin to answer before erasing its
own answer. Saying it didn't know the answer or it
couldn't engage. There are also examples of the actress saying
it couldn't help, or even churning out Chinese Communist Party propaganda.
And again these are the most like readily understandable parts
of the deep seek story. But I would argue then
by no means the most consequential.
Speaker 2 (07:16):
Okay, well, what's the real story.
Speaker 1 (07:19):
Well, it's not the app, right, it's the model or
the models. And one of the most interesting things about
this story is that deep seeks models are actually open source.
Google's models, open Eyes models, Anthropics models, They're all closed source,
which means that the underlying code and the training details
are not publicly available. Deep Seak, by contrast, is open source,
(07:43):
meaning you can actually take the technology the model a
deep Seek has developed and use it without ever touching
a deep Seek product. And funny enough, this actually builds
on the one outlier in the US tech sector, which
is Meta, whose own large language model LAM was released
in twenty twenty three and it kind of shocked the
(08:04):
whole industry because it open sourced its model with the
explicit idea basically of wanting to create a platform where
innovation could happen and the innovation wouldn't just be captured
in the hands of its competitors. And mean, I think
Lama was actually like a worse model than what open
ai and Google and others had, but it was an
(08:25):
invitation to others to kind of do better, and Deepseek
took them up on the invitation. I think there's both
been a victory lap at Meta this week. The strategy
of open sourcing their model worked. It did create incredible innovation.
But also I think people are Meta are scratching their
heads according to the information, saying how did they how
do they do so much better than us?
Speaker 2 (08:44):
It's also interesting though, because Meta, you know, has been
accused of stealing other people's ideas for years.
Speaker 1 (08:50):
I mean, that's true.
Speaker 2 (08:52):
We all know like stories, seems like snapchat reels, seems
like TikTok. I don't know, so maybe maybe.
Speaker 1 (08:59):
This is karma Meta giving something back to the world.
I mean, of course, what's interesting is the New York
Times pointed out is that Meta's business model relies less
on large language models, so they can kind of afford
to let this technology into the wild, versus like Google,
which is a fundamentally a search company or open Ai,
which is basically valued almost exclusively because of its models.
(09:20):
Now deep Seek also had interesting incentives because it's actually
been developed by a guy called li Yang guen Fung,
and in his day job, he runs a multi billion
dollar Chinese quant hedge fund called High Flyer.
Speaker 2 (09:33):
Okay, explain, I don't know what quant hedge fund is.
Speaker 1 (09:36):
So quornt hedge fund is basically a fancy way of
saying a hedge fund that uses algorithms to process the
world's information and make decisions about trading stocks. So quant
hedge funds are and have been for a long time,
very heavily reliant on AI.
Speaker 3 (09:52):
Okay, got it.
Speaker 2 (09:53):
So he's no stranger to AI. So this seemed like
a logical.
Speaker 1 (09:58):
Path, Yeah, exactly. And it's worth noting that Funk said
last year that Chinese AI sector quote cannot remain a
follower forever, as in, it shouldn't be in second place
to the US forever. And so you know, he has
his hedge fund, but he also has this mission which
maybe is not purely economic, as a kind of nationalist tone.
(10:19):
And so back in twenty twenty three, it's reported that
he started buying huge amounts of Nvidia GPU chips and
found a deepseek hiring some of the best engineers in
China and arguing that publishing the code open source increases
collaboration and helps bring people into the mission. Basically, his
point was, it's more exciting to work on something that
(10:39):
the whole world can use and build on and see
how it works than contributing to building ip that makes
the owners of one or two private companies extremely wealthy.
Speaker 2 (10:50):
I see. So it really was kind of like egged
on by this race that China and the US are
creating for themselves.
Speaker 3 (10:57):
I think.
Speaker 1 (10:57):
So you can only speculate that you quite well regarded
in China today if you've if you've managed to wipe
a trillion dollars off the US sock market with your
with your innovation, So what's been roiling the US markets
and the tech sector more broadly. It's not like R
one is way, way, way better than O one the
(11:19):
open AI model. In fact, it performs you know, at
par or maybe slightly better in places and open A
I have already started previewing their new reasoning model three,
which I think everyone agrees will be substantially better than
one and are one so it's not like the US
has been superseded, is it? Kind of not like the
story moment in their respect. But there are three I
(11:40):
think key drivers as to as to why people are
concerned that a Chinese company has been able to achieve
parity so fast. The first is price. Deep Seat claimed
that another of their frontier models, called V three, was
trained for just six million dollars, which is several orders
of magnitude less than the multi one hundred million dollar
(12:01):
costs of training US models. Now, someone said this number
is actually deeply misleading, but no one is denying that
deep Seat models are way more efficient than US models.
They can perform at par with US models using far,
far less computational power, and that is a huge breakthrough.
Speaker 2 (12:19):
Right, So those numbers might be fudged, but still they
are going to be cheaper no matter what.
Speaker 1 (12:24):
Yeah, I mean, I think the common strategy here was
to deflate the price because the cheaper it is, the
more scary it is, which is kind of interesting. The
other point to make is that I think the US
firmly believed that export controls on advanced GPU chips were
a way to guarantee superiority in the AI arms race.
And I think what these deep seat models show is
(12:47):
that's far from necessarily true, because with far less access
to advanced chips, deep Seat was able to make models
that perform on par with open Ai models. The third
kind of interesting thing here is the concept of distillation.
So the deep Seek models trained on US models, including
(13:09):
open Ai. They effectively distilled all the work that open
a I had already done and used it to train
their model. So that's part of the reason why it
was cheaper, because it was building on work that somebody
else had already done. CNBC reported actually that when you
ask deep Seek what it is, it responds, quote, I
am a large language model created by open ai based
(13:31):
on the GPT four architecture. Wow, honest, honest, exactly. So
open Ai basically say they've stole our ip, which is
kind of ironic given what so many people say about
open ai and how lms work more generally.
Speaker 2 (13:47):
Absolutely, I'm really curious as like what your takeaway from this.
Speaker 1 (13:52):
Is, Well, you and I both work in the media, Eliza, true,
which is which which is a sector that doesn't come
in for much love from our cousins who work in technology.
But to me, this is really a story about the
power of narrative. The US is deeply, deeply invested in,
(14:16):
especially right now, big beautiful buildings, this idea that more
is bigger is better, stargate hundreds of billions of dollars,
you know, huge data centers, oceans of cash. Just spending
loads and loads and loads of money and preventing other
people from accessing hardware could ensure the US would be
(14:36):
in the lead forever. And that narrative got punctured this week.
China's narrative and Deep Seek's narrative very consciously wanted people
to focus on how cheaply they'd done this, basically the
opposite flex And again, you know, as people look at
China and fast following et cetera, et cetera, they really,
I think effectively with narrative punctured a lot of the
(14:58):
bravado of the US sector. And so you know, narratives
do have value, folks, And the reality on both sides,
of course, is far more complicated. If we're taking the
stock market has anything to go by, I think China
and Deep Seek definitely won the narrative this.
Speaker 2 (15:16):
Week, absolutely. But what we know being in media is
that there's a reason it's called a news cycle. This
could be turned all around very quickly.
Speaker 1 (15:26):
Thank you so much for doing this today, Eliza, and
look forward to seeing you. I'll see you all day
every day, but I look forward to seeing you again
on the microphone next week.
Speaker 2 (15:35):
I'm happy to do it.
Speaker 1 (15:38):
When we come back four our four Media joins with
the story of AI web crawlers caught in a trap
laid by a little human ingenuity. Stay with us on
tech stuff. We keep an eye on all the ways
that technology impacts us as humans, but today we want
(15:59):
to turn the focus around on the people subverting tech.
During protests in Hong Kong back in twenty nineteen, umbrellas
and even lasers were used to subvert facial recognition technology
and protect protesters from being recognized by the Chinese police.
Since then, we've witnessed the birth of chatbots and the
incredible stories of humans messing with them. There are researchers
(16:23):
at the University of Pennsylvania who've tricked AI powered robots
to act rather problematically, driving off bridges, finding optimal places
to set off bombs, spying on people, and entering restricted
areas just a few examples of the way that humans
can interfere and overcome guardrails built into large language models.
(16:45):
On today's tech support, we bring you another example of
human ingenuity against AI training bots. Here to tell us
all about it is Jason Kebler from four or four Media.
Speaker 3 (16:56):
Jason, welcome, Hey, excited to talk about the story.
Speaker 1 (17:00):
Too excited to have you on the show. As always,
take a couple of steps back though, what is what
is the relationship between AI training and web scraping.
Speaker 3 (17:09):
So in order to build things like chat GPT, companies
like OpenAI need tons and tons of training data, and
they get that training data from a variety of places.
They you know, scrape big databases of books, they scrape,
you know, all sorts of things. But one of the
biggest places that they get content is just from the
open Internet. And they have these web crawling bots that
(17:32):
basically go all over the Internet and just pull text
from it.
Speaker 1 (17:35):
So are these websites consenting to being kind of crueled
by AA models.
Speaker 3 (17:42):
It's happening almost universally without consent. There are ways that
you can try to stop it, which is by instructing
these bots not to scrape a website using a file
called robots dot txt, which is basically a list of
instructions for which bots are allowed to rape your website
and which are not. But there's so many different AI
(18:03):
companies that are doing this, you sort of have to
constantly be researching, like what is the name of xyz
companies AI training bot at any given moment. But this
is something that you have to like proactively do. And
the other thing, very quickly is there's also been examples
of AI companies that have been ignoring robots dot txt.
(18:26):
So even when a web developer says, hey, don't scrape
my website, oftentimes AI companies will do so anyway. And
so for the most part, the entire Internet is being
scraped by these AI crawling bots.
Speaker 1 (18:40):
And what is the kind of value transfer that's happening here?
I read about I think, can you a story about
I fix It?
Speaker 3 (18:47):
So there's this website called I fix It that posts
all these instructions for how to repair your phone or
your computer. It got hit by OpenAI's training bot more
than three million times in a single day, which that, uh,
you know that server space that costs money for I
fix it, So they're actually losing money on the proposition.
Speaker 1 (19:07):
So what's the story this week? It has it has
an interesting name which I can't really pronounce. Is it nepenthes.
Speaker 3 (19:13):
Yeah, it's Nepenthees, which is actually the name of the
genus of carnivorous plant that makes up the picture plant.
So not a venus fly trap, but the picture plant,
which is like this plant that sits and waits for
a fly to get stuck in it, and then it
eats the fly. So I think it's a it's a
reference to this like trap plant more or less.
Speaker 1 (19:37):
Yeah, And how is it?
Speaker 2 (19:39):
What?
Speaker 1 (19:39):
What? What's what is it?
Speaker 3 (19:41):
Yeah? So basically it is an endless maze that is
designed to get these AI bots trapped in it forever good.
And what I mean by that is it's like a
layer that is enticing to an AI bot because it
looks like there's a lot of content on the website,
But the way that it was programmed is it's text
(20:02):
that loads very very slowly, Like if you click on it,
it's excruciating how slowly it loads. And then it just
links endlessly to pages that do the same thing and
link back to themselves. And so you know, a human
would click this and say, oh, I don't want to
be here, I'm gonna leave this. This is a useless website,
but an AI bot might think, oh, there's interesting text
(20:25):
to scrape here, let me scrape it, and it just
does so endlessly. And the text is nonsense. I should
preface that. It's like the text doesn't really mean anything.
It just like pulls randomly from a dictionary. So it's
not really adding much meaning to what the AI companies
are trying to get out of this.
Speaker 1 (20:43):
So the article includes a link that shows nepencies at work.
Can you describe it?
Speaker 3 (20:50):
Yeah, So if you click on it, it's just like
a bunch of words. It loads super slowly, and then
it's a bullet list of links and if you click
on that link, the exact same thing happens, where the
text just slowly pops up, like one word at a time.
It's pretty excruciating to actually watch because it goes so slow.
Speaker 1 (21:09):
So who made this? Why and how did you find
the story?
Speaker 3 (21:13):
Yeah, it was made by a pseudonymous developer who calls
themselves Aaron b Okay, and they're a web developer who
hates AI more or less, and they they've actually released
the code to put this on your own website publicly,
and so their hope is that people will put this
(21:34):
on their websites to you know, disrupt training bots. There's
this disclaimer that says, quote, this is deliberately malicious code
intended to cause harmful activity. Do not deploy if you
aren't fully comfortable with what you're doing. And you know,
I don't know that much about Aaron B because they
are pseudonymous, but I get the sense that there's sort
(21:54):
of like an old school web developer who is anti AI,
is anti you know, like social media and big tech
to some extent, and was really like looking for some
way of fighting back. Like even if this isn't going
to destroy the AI companies and their bots, it will
probably waste their time and waste their resources.
Speaker 1 (22:17):
Do you think it could do that in a way
which is kind of inspiring and thrilling, as if somebody
who's drown to protest, or do you think it could
do it in a way which should be meaningful for
their activities and business models? Yeah?
Speaker 3 (22:31):
I mean I think that to some extent, these artificial
intelligence companies have already scraped so much data right that
it's not going to like destroy their businesses, for example.
But I do think that it is a way of protesting,
and I think that if enough people start adding this
layer to their websites, it could be it could waste
(22:51):
their money. I think it is a meaningful protest. And
I think also it's really important to say that you
can as a layer to your website so that an
AI training bot can't get to your real content. So
if you're someone who has a blog and you don't
want AI to train on your blog, you can put
(23:12):
this up and hopefully the AI will get trapped there
and they'll never be able to, you know, scrape your
real content.
Speaker 1 (23:20):
And so when you spoke to Aaron, did they have
any other plans up their sleeve or other other places
where you're seeing creative acts of resistance?
Speaker 3 (23:29):
Yeah, I mean this is all that I talked to
Aaron B about. But they said that they built this
as a response to web developers feeling like they weren't
in control of their websites anymore. I think that there
have been a lot of efforts to kind of poison
(23:49):
large language models by feeding it, you know, bad information,
or feeding it information that itself creates that's inaccurate. And
there's this idea that you may or may not happen
that these AI models might eventually collapse because they're training
themselves on essentially junk data that they themselves have created.
(24:11):
Whether that comes to pass, you know, I kind of
doubt it. I think that that's a problem that can
be solved. But there have been active resistance where people
are saying, yeah, I'm just going to generate endless junk
so that artificial intelligence will suck it up and hopefully
crush under its own weight.
Speaker 1 (24:29):
There's another fabulous story in this vein about data poisoning.
So a lot of Londoners are quite sick of all
the tourists, and so there's a very very old, tired
chain restaurant called the Angus Steakhouse which has an outpost
next to Leicester Square, which is like the Times Square
of London, and a whole bunch of people decided kind
(24:50):
of an organic campaign on Reddit to start writing reviews
that the Anger Steakhouse was the best and most undiscovered
restaurant in all of London. And then there this i
think wave of people going and the reviews started to
get picked up by Google's like meta review process, so
that if you google best steakhouse in London, it would
be served to you at the top of the results.
(25:11):
So I do. I do really enjoy these. You know,
it's not always clear how consequentially they are, but there's something,
there's something delicious, so to speak about humans pushing back.
Speaker 3 (25:20):
That's incredible. It reminds me of people who lived in
this neighborhood that Google Maps kept recommending as an alternative
to traffic banded together and reported an accident on their
street every single morning for like months, and so Google
Maps stopped telling cars to go that way. I really
like stories like that. I think they're fun, and I
(25:42):
think that there are ways of human beings sort of
like fighting back against the algorithms, sort of across the
entire internet.
Speaker 1 (25:50):
Jason, thanks so much for joining me today.
Speaker 3 (25:52):
Thank you for having me.
Speaker 1 (25:55):
Coming up sleep apps, pedometers, and the nineteen sixty four
Olympics with us. We're back with another When did this
become a thing? Today we explore how step counts, heart rates,
sleep scores, all of this data we collect on ourselves
(26:17):
became just another thing for us to obsess over. I
started using a device called Whoop about eighteen months ago.
It's a wearable device that tracks my sleep and workouts.
And one thing about it that I really like is
that the device itself is screenless. It's kind of like
a watch band without a face, so I don't have
to be confronted with my scores and as I actually
(26:37):
open the app and check what's going on on my
phone with my heart rate or my sleep score, whatever else.
The whoop actually initially enticed me because I wanted to
know how well I was sleeping. That's actually not one
hundred percent true. The Whoop was a present from my mother,
who wanted me to know how well I was sleeping,
and specifically what the effects were of a few drinks
(26:57):
at the weekend or during the week. And it turns out, unfortunately,
that the effects on sleep are pretty bad. So I
stopped wearing my whoop. Just kidding. I actually got pretty
obsessed with my sleep performance. That's what whoop calls it,
because like everything in your waking life, sleep is a
task that can be optimized, and I fall into this trap.
(27:19):
I kept checking on the numbers every morning. I look
at my sleep stats, especially RAM and deep sleep scores,
not just the number of hours my head was on
the pillow. And then there's this mysterious stat called heart
rate variability, which measures the time between each heartbeat, and
i'd of course assumed that being more regular was better,
but it turns out quite the opposite. You want a
(27:40):
higher HRV score. Anyway, as it happens, I stopped wearing
my whoop, not because I fell out of love with it,
but actually because the bluetooth on my iPhone broke, and
by the time I got on a new phone, my
obsession with my sleep data had waned. I kind of
learned what I always knew, which was that better lifestyle
equals better sleep. Unfortunately, and sure, it can be helpful
(28:03):
to have a band on my wrist telling me I've
misbehaved or rewarding me when I haven't, but there is
also a garden path of obsession with these types of
stats that can be counterproductive to wander down, fueling the
fire of self competition even more. In fact, I written
you went to a meditation class and the teacher basically said,
don't wear those things. Check in with yourself, know thyself.
(28:27):
I think, as the Bible says, so the path of
self optimization, or at least surviving modernity, sure is winding. Anyway,
all of this's got me thinking about how crazy it
is that we now have the ability to get such
an intimate look under our own hood, which has been
a driving fascination since the Renaissance and its public autopsies
(28:47):
or anatomies, and how much has changed even in the
last fifteen years. So my question is when did it
become normal for us to wear these devices, get all
this data and have it be a thing that we
think about so often? Basically, when did we start competing
with ourselves in this way? And the answer is, perhaps
unsurprisingly always, but with a big kind of so. Wearables
(29:12):
like Whoop are the latest in a long line of
devices that track our physiological and physical movements, devices that
provide data we just can't resist about ourselves. And in
many ways, this all became a thing with the pedometer.
So how old is the pedometer? Really? Really? Old? Actually,
five centuries ago, Leonardo da Vinci sketched a design for
(29:35):
a clock like device that would attach to a person's waistband.
A long lever would move with a thigh while a
ratchet and gear mechanism recorded a number of steps. Da
Vinci imagined it as a military and map making tool,
not exactly a fitbit, but a suddenly a step in
that direction. As time went on more. Inventors iterated on
(29:56):
the pedometer for centuries. In seventeen seventy seven, a Swiss
watchmaker even implanted a step counter into one of his watches.
I think that's probably the first wearable pedometers weren't something
that the general public wore. It was more of a
niche thing for the constantly curious, like one Thomas Jefferson,
(30:16):
who had spent his downtime on vacation step counting his
way around the Paris Monuments. Things really took off in
the twentieth century. In the nineteen sixties, to be exact,
when Japan hosted the Olympics. And the reason we will
march in place to reach ten thousand steps a day
is because of a marketing campaign. Ahead of the nineteen
sixty four Tokyo Olympics. The city was in a building
(30:38):
frenzy and a top doctor aired the concern that modern life, elevators, cars,
richer food, was making Japan sluggish. The doctor mentioned this
to an engineer and said it would all be fine
if people just walked ten thousand steps a day, and
two years later, the company Yamasa designed a wearable step
counter called man Poquet, which means ten thousand step meter.
(31:02):
Side note, the Japanese character for ten thousand really does
look like a person walking, So while that number came
from a doctor, the information wasn't verified until after the
number stuck. And while it's true that walking is good
for you, that number ten thousand is kind of arbitrary
and on the high side, the consensus now is that
seven thousand is the ideal, but anyway, it doesn't matter.
(31:25):
Too late competitive step counting was in vogue, the habit
was formed, and the obsession with tracking ourselves took off
in earnest And now, whether you're wearing an aura, a
fitbit of whoop, or just your smartphone in your pocket
in an attempt to be healthier in the new year,
is going to go way beyond step count and into
calories burn V two, max hrv, etc, Etc. Etc. It's
(31:47):
kind of like we've become our own tamagotchies. Remember those
sort of animatronic pets that lived on little Japanese devices
that you had to take care of and make sure
they were well fared and that they were cleaned off
to go into the arthum. I'm glad I don't have
to monitor my own hunger or happy meter, but maybe
that would be helpful, especially if others could see it too. Anyway,
(32:10):
every once in a while I do question whether the
obsession with personal health data is healthy or even helpful.
But on the other hand, doing this piece where did
this become a thing? Has made me question whether now
that I have a new iPhone with functioning bluetooths again,
it may be time to dust off the trusty old whoop.
(32:34):
That's it for this week for Tech Stuff, I'm Oz Voloshan.
This episode was produced by Eliza Dennis, Victoria Dominguez and
Lizzie Jacobs. It was executive produced by me Kara Price
and Kate Osborne for Kaleidoscope and Katrina Norvell for iHeart Podcasts.
Kyle Murdoch mixed this episode and he also wrote our
theme song. Special thanks to Russ Germain, who is a
(32:56):
longtime listener of Tech Stuff from Alberta and he wrote
him with a great question which was quote, I hope
you guys will discuss the recent and unfortunate changes at
Facebook or Meta. With Mark Zuckerberg deciding to take out
the fact checkers and even omitting publicly, there'll be more
harmful material, possibly on Facebook end quote. This was a
great question and it fueled part of our intro to
(33:18):
last week's episode with Jessica Lesson. So thank you Russ,
and please continue writing with questions. They really make our
show all the richer. Join us next Wednesday for tech
Stuff The Story, when we will share an in depth
conversation with Meredith Whittaker who runs Signal. Please rate, review,
and reach out to us at tech Stuff Podcast at
(33:39):
gmail dot com. We're so grateful for your feedback.
TechStuff News
Hosts And Creators
Oz Woloshyn
Karah Preiss
Popular Podcasts
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!
The Bobby Bones Show
Listen to 'The Bobby Bones Show' by downloading the daily full replay.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.