August 23, 2024 • 20 mins
Meta and Snap each plan to introduce new AR glasses, but you probably won't be able to get your hands on them. Plus, Cruise strikes up a partnership with Uber, Google introduces a photo editing feature that muddies the waters and an enterprising hacker attempts to fake his own death. Through hacking. Plus more!
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 (00:04):
Welcome to tech Stuff, a production from iHeartRadio. Hey there,
and welcome to tech Stuff. I'm your host, Jonathan Strickland.
I'm an executive producer with iHeart Podcasts and How the
tech are you. It's time for the tech news for
the weekending Friday, August twenty third, twenty twenty four. And
(00:25):
first up, we have what I guess you could call
non news news. So recently online tech news outlets began
to report that Google was essentially killing off the Fitbit brand.
Google acquired Fitbit back in twenty nineteen, and over time,
the company has incorporated more of the same sort of
tech found and Fitbit activity trackers into the Google branded
(00:48):
smart watches. So the reporting seemed to suggest that Google
was going to sunset the Fitbit brand, or at least
reduce Fitbit to just a few models of its fitness trackers,
while the fit Bit branded smart watches like the Sense
and the Versa would essentially go away. However, Sharon Harding
(01:09):
of ours Tetnica followed up on this. She contacted Google
directly and asked if, in fact, the company was going
to reduce the fit Bit line to activity trackers like
the Inspire and the charge models. The rep from Google
said the earlier reporting was inaccurate, that Google, in fact
had just released a fit Bit branded smart watch for kids.
(01:31):
This is like a tongue twister, and I'm terrible at
those anyways. Harding points out the concerns about Google killing
off product lines. That's understandable because Google has a huge
body count when it comes to products and services that
it once launched that have long since pushed up the daisies.
But it seems like Fitbit is not going to be
one of those, at least according to company representatives. The
(01:54):
vergess Lauren Finer has an article titled Google sales reps
allegedly keep telling advertisers how to target teams. So this
relates to a story I talked about recently on Tech Stuff.
Google follows an industry practice that, in theory at least
means the company does not engage in targeted or personalized
(02:14):
advertising for any user under the age of eighteen. However,
according to multiple sources, now Google has made use of
a bit of a loophole. The company can target unknown users.
So these are users who do not have age information
built into their profiles, like they have not indicated what
their age is to Google, so there's no confirmation one
(02:37):
way or the other to say the user is above
or below the age of eighteen. So you could say
this is kind of a case of plausible deniability. But
as I'm sure you're well aware, it really does not
take much work to get a general feel for someone's
age based upon their web activity. So with even a
relatively simple data analysis, pass can start putting folks into
(03:01):
various buckets, including by age range. So while technically the
unknown users aren't registered as teens, you can effectively target
teens with advertising through this kind of roundabout approach. Finer
cites an article on Adweek and another in Financial Times
indicating that this is an issue that multiple outlets have
(03:23):
looked at recently. Google's response includes a statement from Josell Booth,
a spokesperson for the company, who categorically emphasized that Google's
policy is not to personalize advertising for anyone under the
age of eighteen, regardless of whether it's through direct data
or inferring age from supporting data, and that the company
would stress this to sales reps so that you know
(03:44):
they stop suggesting it to potential clients. That Google can
you know totally get advertisers linked up to impressionable teenagers
to knock that stuff off. Alex Heath, also of The Verge,
has a piece that lets us know reality is about
to get a bit more augmented. That is, both Snap
and Meta have plans to unveil AR glasses in the
(04:05):
upcoming weeks. In Snap's case, this would be the latest
version of the company's Spectacles product. This would be generation
number five for those of y'all keeping count. According to Heath,
Snap will be showing the soft terraing Snap's partner summit
on September seventeenth, Then just a week later, Mark Zuckerberg
plans to unveil Meta's own AR glasses, which are code
(04:27):
named Orion. However, Heath's sources tell him that in neither
case will these products ever hit the consumer market. These
are more like Microsoft's HoloLens in that regard. These pieces
of hardware are really meant to give developers a platform
to build upon. The issue here is that while the
potential for AR seems pretty darn limitless, the truth is
(04:50):
very few apps have been built that leverage AR in
a way that makes it the best tool for the job.
As we have seen with Apple's vision headset. Having great
hardware where isn't necessarily enough. You need the applications to
be there too. So while we should be seeing some
impressive demonstrations of this technology, assuming everything goes as planned,
we won't actually be using this stuff anytime soon. Heath
(05:15):
says that Snap will have around ten thousand units produced
and Meta will have even fewer than that, and I'm
sure that won't stop some tech enthusiasts from trying to
get their hands on these things, despite the fact that
there just isn't much you can do with them yet.
Some folks just have an almost pathological need to have
the latest technology. I should know. I used to be
one of them. Time to talk about AI for a
(05:36):
good long while, because I mean, it's twenty twenty four, y'all.
So first up, multiple news outlets have reported that Meta
has rolled out a couple of new web crawling bots
designed to gather data for the purposes of training AI models,
and further that these bots ignore attempts to block them,
so web page builders, in case you've never done this before,
(05:56):
they have the option to include a little line of
text that essentially tells bots to buzz off, so you
might want to do that if you do not want
your web page indexed for like search purposes, And you
might want to do that if you only wanted authorized
individuals to even know about the page in the first place.
But these bots reportedly ignore these kinds of lines of text,
(06:19):
and they will crawl a site even if the web
administrators said, please don't index the site. And I find
that really interesting because Meta very much takes the stance
that crawling sites like Facebook and the like is expressly
against their rules, that no one is supposed to treat
Meta that way, and yet here they are producing bots
(06:41):
that are apparently engaging in the very same behavior that
Meta prohibits on its owned and operated sites. How about
that anyway, This is really all part of the AI
arms race, where various AI companies are desperate to get
ever larger pools of data in order to train their
large language models and make the next AI tool guaranteed
(07:02):
to create massive ethical problems. Web administrators can end up
in a pickle when companies like Meta and Google engage
in this kind of activity, because often it means that
if you are successful in blocking the AI crawlers, it
means you're also having to block the index bots, which
means your site is not going to pop up in
like search results and such. So website operators they're pressured
(07:27):
to allow this AI crawling activity or else potentially miss
out on being discoverable on these massive platforms. And it
doesn't do you much good if you built something and
no one knows about it, right, So it becomes this
double edged sword. It could be like, well, yeah, I
want to be in search because I want people to
be able to find me, but I don't want it
(07:48):
to be crawled for the purposes of AI. Well, when
it's the same companies doing both, that becomes a problem,
and it's kind of like extortion if you think about it.
It's almost as if some of these big companies act
in ways that can be a bit anti competitive. Folks
over at the University of Texas have developed an earthquake
detection tool that uses AI to look for signals that
could indicate an upcoming earthquake, and the results have been promising.
(08:12):
According to SyTech Daily, the researchers achieved a seventy percent
accuracy rating in predicting earthquakes a week before the earthquakes
actually happened. The research was conducted in China. I think
this is pretty darn cool, but we do need to
remember this is by no means a perfect tool. It's
got a long way to go. But according to the researchers,
(08:34):
there were eight false positives, meaning the tool predicted an
earthquake but nothing actually happened. That's an issue. Also, the
predictions weren't exactly laser precise. According to the article, the
fourteen earthquake predictions that the researchers counted as successes were
within two hundred miles of an actual earthquake. That does
(08:55):
make one wonder if the tool was successful at predicting
the earthquake, or maybe it was just a matter of coincidence. However,
according to the article, the predicted strength of the earthquakes
was very close to what actually happened. That makes me
more inclined to think that this is not just coincidence.
It's one thing to say an earthquake is going to
happen on this day, generally around this time, and at
(09:16):
about this strength. If you were only getting like an
earthquake happening within two hundred miles of where you predicted,
but the strength wasn't anywhere close to what you predicted
that to me would feel like coincidence. Getting the strength
just about right. That seems to whittle that down a bit,
but it does mean there's a limitation as to how
precise this tool can be when it comes to locating
where an earthquake is going to happen. The research obviously
(09:39):
needs to continue. A lot more work needs to be
done in order to turn this into a really useful technology.
As it stands, sending out a warning a week ahead
of time that someone might be within a two hundred
mile radius of a future earthquake, that seems limited in
its usefulness unless we're talking about like a real whopper
of an earthquake, in which case it could potentially help
save countless lives, assuming that people actually heeded the warning. Okay,
(10:04):
we've got a lot more to talk about in today's news,
but first let's take a quick break to thank our sponsors.
We're back and we're headed back to the Verge. Last
week it was all ours Technica. This week gets the Verge. Anyway.
(10:26):
There are a pair of articles covering the same general
topic over on the Verge. One is by Alison Johnson,
the others by Sarah Jong, but they're both about Google's
Reimagine function in the new Google Pixel nine smartphones. So
this feature allows you to make some pretty massive changes
to photos that you've already taken, and you can use
text based prompts to have AI alter those images in
(10:49):
various ways. So one example that the articles used was
they showed a photo of just a street, just a
normal street, and then a subsequent text edit added in
a massive pothole incorporated into that street, and sure enough,
it was an edited photo that looked very convincing, like
like the pothole was actually there. Both Johnson and to
(11:11):
a greater extent, Johng point out that the feature allows
for fakery and image manipulation on a grand scale. Once
upon a time, you needed at least to be proficient
with tools like Photoshop to manipulate images convincingly, and even
then there were like telltale signs that some altering had happened.
But now AI takes care of all of this for you,
(11:32):
and it can be pretty darn good at fooling folks. Now,
the AI is supposed to have guardrails that are meant
to prevent users from doing really awful stuff, like you
wouldn't want someone to have a photo and then use
AI to just litter the ground with like dead puppies
or something that would be horrifying. But the folks at
the Verge found they were able to insert a great
(11:54):
deal of troubling imagery into photographs just by adding some
creative thinking to get around the guardrails. While being direct
and blunt in your text directions might result in a
denial saying no, that's against the policy or whatever, if
you're a little more circumspect, you can often get the
same results. And so the Verge showed off images that
appear to portray such disturbing scenes as a collision that
(12:16):
happened between a car and a bicycle on a city street,
or images where there appears to be a body laying
underneath a bloody sheet on the ground. It's not exactly
the most positive showcase for an image manipulation tool using AI. Moreover,
even if we remove the obvious cases of like unintended consequences,
the end result is that this tool means seeing is
(12:39):
absolutely not believing. When image manipulation is so easy that
anyone with the right kind of smartphone can do it
with no training needed, what does that mean for information?
How do we know what to trust. How could such
a tool be used to deceive others, either just for
kicks or for personal gain or whatever. Are the benefits
(13:01):
of this technology such that they actually outweigh the risks.
Then there's also the element of the liar's dividend. This
is the defense that someone who is absolutely guilty of
something could use. They could say, Oh, sure it looks
like that was me robbing that convenience store, but that's
clearly an AI altered image. I'm innocent. That's the liar's dividend.
(13:22):
Making matters worse is that Reimagine, at least currently doesn't
apply a digital watermark to altered images like purely AI
generated images often have a water mark, but not these.
Johnson points out that the meta data for the image
includes a record that it was edited through Reimagine, But
she also points out that you can get around that
(13:43):
just by taking a screenshot of the photo in question.
That just strips out all the metadata, because now you
just have a picture of a picture and there's no
record there that Reimagine was used to alter it. Blah.
You might remember that way back when the twenty twenty
four Democratic primary were going on, which seems like it
happened in a different world at this point. But you
(14:04):
might remember there were reports of an AI generated voice
that was impersonating US President Joe Biden, and it was
going out to potential voters in New Hampshire, and the
voice was urging voters to just stay home and not
go and vote in the primaries. Well, now the Federal
Communications Commission, or FCC has ordered the telecom company Lingo
(14:24):
Telecom to pay a one million dollar civil penalty for
allowing those calls to go out over its network. Now,
to be clear, Lingo Telecom wasn't responsible for creating those calls.
That honor falls to a political consultant named Steve Kramer,
who in turn was working on behalf of a candidate
named Dean Phillips who was running an opposition to Joe Biden.
(14:47):
But Lingo Telecom allowed the calls to go over its network,
which the FCC deemed as a violation of the know
your Customer and Know your Provider sets of rules. Kramer
is also facing a fine could be up to around
six million dollars. These penalties are meant to send a
message to folks who are considering a similar scheme that
if you do this kind of thing, it's going to
(15:07):
cost you. As for Lingo, the company also agreed to
make some changes to how it operates to WITT, weeding
out spoofed phone numbers and only presenting a number when
Lingo can verify that it's exactly where a call is
really coming from. Presumably you would otherwise see something like
unknown caller or something like that on your caller ID.
(15:28):
Lingo must also verify the identities of customers and work
with upstream providers that have quote unquote robust robocall mitigation.
So yeah, this is really sitting a message of saying
this kind of thing will not be tolerated. In twenty
twenty three, GM's cruise business shut down effectively after one
of its autonomous robotaxis dragged the pedestrian for twenty feet
(15:51):
in San Francisco before it came to a stop. The
pedestrian had already been struck by another car that one
was operated by a human. Pretty awful, like a really
hard ruble sequence of events, and CRUZ faced a massive investigation.
The CEO of the division promptly jumped ship, as did
several other leaders, and GM laid off a significant number
(16:11):
of workers within the Cruz division. But now Cruz is
back in the news, having struck a partnership deal with Uber.
New CEO Mark Witten said, quote, we are excited to
partner with Uber to bring the benefits of safe, reliable
autonomous driving to even more people, unlocking a new era
of urban mobility end quote. Which is interesting to me.
(16:34):
Uber itself really was aggressively pursuing robotaxi strategies several years ago,
because I mean, cutting human drivers out of the equation
means more money for the home office. Am I right?
But snarky comments aside. Uber pretty much pulled the plug
on its own efforts after a tragic incident in twenty
eighteen involving another pedestrian accident. Uber then switched to partnering
(16:57):
with companies in the autonomous vehicle space rather than pursuing
their own program. So can these two companies, each with
blemishes on their respective records, team up to create something
that's safe and reliable. Cruz is currently conducting autonomous vehicle
testing with supervising safety drivers in cities like Houston and Phoenix.
No word yet on when those driver lest Uber rides
(17:19):
will become a reality, or specifically which markets that might
happen in I know what you're thinking, you know, Jonathan,
it's been a hot minute since I've learned about a
new streaming video service launching. Well, rumor has it that
we might be getting yet another one from a seemingly
unlikely source. That source is Chick fil A, the fast
food restaurant known for chicken, among other things. Deadlines Peter
(17:41):
White reports that Chick fil A is planning a service
centered primarily around reality television and unscripted content and game
show programming, all with like a family friendly focus. Presumably
the programming on this service would in some way advertise
or promote the company, perhaps through the production of branded content.
How many folks are out there itching to sign up
to yet another streaming service, let alone one spearheaded by
(18:04):
a restaurant company. I have no clue. I'm not going
to write it off just yet, as it could always
surprise me. But my first impression is this is going
to be a very tough sell, particularly during a time
when people are already taking a harder look at their
family budgets for stuff like entertainment. Jesse Kiff is in
hot water for hacking into a government registry in Hawaii
for what purpose? Faking his own death. Kiff hacked into
(18:26):
the Hawaii death registry system and marked himself down as
previously alive or no longer breathing, or debt as a doornail.
To be honest, I don't know what the checkboxes actually say,
but the point is Kiff was faking his own debt.
He was also using fake credentials in an effort to
secure a credit card or debit a card account. It's
awfully hard to navigate the modern world unless you've got
(18:47):
access to that cheddar. So why was Kiff doing all this? Well,
apparently it was in order to avoid having to make
child support payments. He's already been tried and found guilty.
He faces a prison term of sixty nine months. Nice
he'll have to serve eighty five percent of that sentence,
after which he will be released, but will remain under
supervision for three years. Now for some recommended reading, So
(19:08):
first a recommend checking out Eric Berger's piece for Ours
Tetnica titled Against All Odds and Asteroid Mining Company appears
to be making headway, which is a cool story. We've
been hearing about potentials for asteroid mining for several years now,
so it's neat getting an update. Next up, Patrick George
has a piece in the Atlantic titled The Hardest Sell
in American Car Culture, and it's about how the Ford
(19:29):
Car Company wants to encourage American car shoppers to think
about smaller vehicles rather than the trucks and SUVs the
industry has kind of migrated to in the US over
the previous years. And that's because smaller cars are lighter.
Lighter cars are easier to move, and that means the
battery requirements for EV's that are smaller are more manageable
than for those big old chonkers that are currently favored
(19:51):
by the US. So if the US is to move
to more evs, part of the picture may also mean
driving smaller vehicles. That's it for this week. I hope
you are all well and I'll talk to you again
really soon. Tech Stuff is an iHeartRadio production. For more
(20:13):
podcasts from iHeartRadio, visit the iHeartRadio app, Apple Podcasts, or
wherever you listen to your favorite shows.
Welcome to tech Stuff, a production from iHeartRadio. Hey there,
and welcome to tech Stuff. I'm your host, Jonathan Strickland.
I'm an executive producer with iHeart Podcasts and How the
tech are you. It's time for the tech news for
the weekending Friday, August twenty third, twenty twenty four. And
(00:25):
first up, we have what I guess you could call
non news news. So recently online tech news outlets began
to report that Google was essentially killing off the Fitbit brand.
Google acquired Fitbit back in twenty nineteen, and over time,
the company has incorporated more of the same sort of
tech found and Fitbit activity trackers into the Google branded
(00:48):
smart watches. So the reporting seemed to suggest that Google
was going to sunset the Fitbit brand, or at least
reduce Fitbit to just a few models of its fitness trackers,
while the fit Bit branded smart watches like the Sense
and the Versa would essentially go away. However, Sharon Harding
(01:09):
of ours Tetnica followed up on this. She contacted Google
directly and asked if, in fact, the company was going
to reduce the fit Bit line to activity trackers like
the Inspire and the charge models. The rep from Google
said the earlier reporting was inaccurate, that Google, in fact
had just released a fit Bit branded smart watch for kids.
(01:31):
This is like a tongue twister, and I'm terrible at
those anyways. Harding points out the concerns about Google killing
off product lines. That's understandable because Google has a huge
body count when it comes to products and services that
it once launched that have long since pushed up the daisies.
But it seems like Fitbit is not going to be
one of those, at least according to company representatives. The
(01:54):
vergess Lauren Finer has an article titled Google sales reps
allegedly keep telling advertisers how to target teams. So this
relates to a story I talked about recently on Tech Stuff.
Google follows an industry practice that, in theory at least
means the company does not engage in targeted or personalized
(02:14):
advertising for any user under the age of eighteen. However,
according to multiple sources, now Google has made use of
a bit of a loophole. The company can target unknown users.
So these are users who do not have age information
built into their profiles, like they have not indicated what
their age is to Google, so there's no confirmation one
(02:37):
way or the other to say the user is above
or below the age of eighteen. So you could say
this is kind of a case of plausible deniability. But
as I'm sure you're well aware, it really does not
take much work to get a general feel for someone's
age based upon their web activity. So with even a
relatively simple data analysis, pass can start putting folks into
(03:01):
various buckets, including by age range. So while technically the
unknown users aren't registered as teens, you can effectively target
teens with advertising through this kind of roundabout approach. Finer
cites an article on Adweek and another in Financial Times
indicating that this is an issue that multiple outlets have
(03:23):
looked at recently. Google's response includes a statement from Josell Booth,
a spokesperson for the company, who categorically emphasized that Google's
policy is not to personalize advertising for anyone under the
age of eighteen, regardless of whether it's through direct data
or inferring age from supporting data, and that the company
would stress this to sales reps so that you know
(03:44):
they stop suggesting it to potential clients. That Google can
you know totally get advertisers linked up to impressionable teenagers
to knock that stuff off. Alex Heath, also of The Verge,
has a piece that lets us know reality is about
to get a bit more augmented. That is, both Snap
and Meta have plans to unveil AR glasses in the
(04:05):
upcoming weeks. In Snap's case, this would be the latest
version of the company's Spectacles product. This would be generation
number five for those of y'all keeping count. According to Heath,
Snap will be showing the soft terraing Snap's partner summit
on September seventeenth, Then just a week later, Mark Zuckerberg
plans to unveil Meta's own AR glasses, which are code
(04:27):
named Orion. However, Heath's sources tell him that in neither
case will these products ever hit the consumer market. These
are more like Microsoft's HoloLens in that regard. These pieces
of hardware are really meant to give developers a platform
to build upon. The issue here is that while the
potential for AR seems pretty darn limitless, the truth is
(04:50):
very few apps have been built that leverage AR in
a way that makes it the best tool for the job.
As we have seen with Apple's vision headset. Having great
hardware where isn't necessarily enough. You need the applications to
be there too. So while we should be seeing some
impressive demonstrations of this technology, assuming everything goes as planned,
we won't actually be using this stuff anytime soon. Heath
(05:15):
says that Snap will have around ten thousand units produced
and Meta will have even fewer than that, and I'm
sure that won't stop some tech enthusiasts from trying to
get their hands on these things, despite the fact that
there just isn't much you can do with them yet.
Some folks just have an almost pathological need to have
the latest technology. I should know. I used to be
one of them. Time to talk about AI for a
(05:36):
good long while, because I mean, it's twenty twenty four, y'all.
So first up, multiple news outlets have reported that Meta
has rolled out a couple of new web crawling bots
designed to gather data for the purposes of training AI models,
and further that these bots ignore attempts to block them,
so web page builders, in case you've never done this before,
(05:56):
they have the option to include a little line of
text that essentially tells bots to buzz off, so you
might want to do that if you do not want
your web page indexed for like search purposes, And you
might want to do that if you only wanted authorized
individuals to even know about the page in the first place.
But these bots reportedly ignore these kinds of lines of text,
(06:19):
and they will crawl a site even if the web
administrators said, please don't index the site. And I find
that really interesting because Meta very much takes the stance
that crawling sites like Facebook and the like is expressly
against their rules, that no one is supposed to treat
Meta that way, and yet here they are producing bots
(06:41):
that are apparently engaging in the very same behavior that
Meta prohibits on its owned and operated sites. How about
that anyway, This is really all part of the AI
arms race, where various AI companies are desperate to get
ever larger pools of data in order to train their
large language models and make the next AI tool guaranteed
(07:02):
to create massive ethical problems. Web administrators can end up
in a pickle when companies like Meta and Google engage
in this kind of activity, because often it means that
if you are successful in blocking the AI crawlers, it
means you're also having to block the index bots, which
means your site is not going to pop up in
like search results and such. So website operators they're pressured
(07:27):
to allow this AI crawling activity or else potentially miss
out on being discoverable on these massive platforms. And it
doesn't do you much good if you built something and
no one knows about it, right, So it becomes this
double edged sword. It could be like, well, yeah, I
want to be in search because I want people to
be able to find me, but I don't want it
(07:48):
to be crawled for the purposes of AI. Well, when
it's the same companies doing both, that becomes a problem,
and it's kind of like extortion if you think about it.
It's almost as if some of these big companies act
in ways that can be a bit anti competitive. Folks
over at the University of Texas have developed an earthquake
detection tool that uses AI to look for signals that
could indicate an upcoming earthquake, and the results have been promising.
(08:12):
According to SyTech Daily, the researchers achieved a seventy percent
accuracy rating in predicting earthquakes a week before the earthquakes
actually happened. The research was conducted in China. I think
this is pretty darn cool, but we do need to
remember this is by no means a perfect tool. It's
got a long way to go. But according to the researchers,
(08:34):
there were eight false positives, meaning the tool predicted an
earthquake but nothing actually happened. That's an issue. Also, the
predictions weren't exactly laser precise. According to the article, the
fourteen earthquake predictions that the researchers counted as successes were
within two hundred miles of an actual earthquake. That does
(08:55):
make one wonder if the tool was successful at predicting
the earthquake, or maybe it was just a matter of coincidence. However,
according to the article, the predicted strength of the earthquakes
was very close to what actually happened. That makes me
more inclined to think that this is not just coincidence.
It's one thing to say an earthquake is going to
happen on this day, generally around this time, and at
(09:16):
about this strength. If you were only getting like an
earthquake happening within two hundred miles of where you predicted,
but the strength wasn't anywhere close to what you predicted
that to me would feel like coincidence. Getting the strength
just about right. That seems to whittle that down a bit,
but it does mean there's a limitation as to how
precise this tool can be when it comes to locating
where an earthquake is going to happen. The research obviously
(09:39):
needs to continue. A lot more work needs to be
done in order to turn this into a really useful technology.
As it stands, sending out a warning a week ahead
of time that someone might be within a two hundred
mile radius of a future earthquake, that seems limited in
its usefulness unless we're talking about like a real whopper
of an earthquake, in which case it could potentially help
save countless lives, assuming that people actually heeded the warning. Okay,
(10:04):
we've got a lot more to talk about in today's news,
but first let's take a quick break to thank our sponsors.
We're back and we're headed back to the Verge. Last
week it was all ours Technica. This week gets the Verge. Anyway.
(10:26):
There are a pair of articles covering the same general
topic over on the Verge. One is by Alison Johnson,
the others by Sarah Jong, but they're both about Google's
Reimagine function in the new Google Pixel nine smartphones. So
this feature allows you to make some pretty massive changes
to photos that you've already taken, and you can use
text based prompts to have AI alter those images in
(10:49):
various ways. So one example that the articles used was
they showed a photo of just a street, just a
normal street, and then a subsequent text edit added in
a massive pothole incorporated into that street, and sure enough,
it was an edited photo that looked very convincing, like
like the pothole was actually there. Both Johnson and to
(11:11):
a greater extent, Johng point out that the feature allows
for fakery and image manipulation on a grand scale. Once
upon a time, you needed at least to be proficient
with tools like Photoshop to manipulate images convincingly, and even
then there were like telltale signs that some altering had happened.
But now AI takes care of all of this for you,
(11:32):
and it can be pretty darn good at fooling folks. Now,
the AI is supposed to have guardrails that are meant
to prevent users from doing really awful stuff, like you
wouldn't want someone to have a photo and then use
AI to just litter the ground with like dead puppies
or something that would be horrifying. But the folks at
the Verge found they were able to insert a great
(11:54):
deal of troubling imagery into photographs just by adding some
creative thinking to get around the guardrails. While being direct
and blunt in your text directions might result in a
denial saying no, that's against the policy or whatever, if
you're a little more circumspect, you can often get the
same results. And so the Verge showed off images that
appear to portray such disturbing scenes as a collision that
(12:16):
happened between a car and a bicycle on a city street,
or images where there appears to be a body laying
underneath a bloody sheet on the ground. It's not exactly
the most positive showcase for an image manipulation tool using AI. Moreover,
even if we remove the obvious cases of like unintended consequences,
the end result is that this tool means seeing is
(12:39):
absolutely not believing. When image manipulation is so easy that
anyone with the right kind of smartphone can do it
with no training needed, what does that mean for information?
How do we know what to trust. How could such
a tool be used to deceive others, either just for
kicks or for personal gain or whatever. Are the benefits
(13:01):
of this technology such that they actually outweigh the risks.
Then there's also the element of the liar's dividend. This
is the defense that someone who is absolutely guilty of
something could use. They could say, Oh, sure it looks
like that was me robbing that convenience store, but that's
clearly an AI altered image. I'm innocent. That's the liar's dividend.
(13:22):
Making matters worse is that Reimagine, at least currently doesn't
apply a digital watermark to altered images like purely AI
generated images often have a water mark, but not these.
Johnson points out that the meta data for the image
includes a record that it was edited through Reimagine, But
she also points out that you can get around that
(13:43):
just by taking a screenshot of the photo in question.
That just strips out all the metadata, because now you
just have a picture of a picture and there's no
record there that Reimagine was used to alter it. Blah.
You might remember that way back when the twenty twenty
four Democratic primary were going on, which seems like it
happened in a different world at this point. But you
(14:04):
might remember there were reports of an AI generated voice
that was impersonating US President Joe Biden, and it was
going out to potential voters in New Hampshire, and the
voice was urging voters to just stay home and not
go and vote in the primaries. Well, now the Federal
Communications Commission, or FCC has ordered the telecom company Lingo
(14:24):
Telecom to pay a one million dollar civil penalty for
allowing those calls to go out over its network. Now,
to be clear, Lingo Telecom wasn't responsible for creating those calls.
That honor falls to a political consultant named Steve Kramer,
who in turn was working on behalf of a candidate
named Dean Phillips who was running an opposition to Joe Biden.
(14:47):
But Lingo Telecom allowed the calls to go over its network,
which the FCC deemed as a violation of the know
your Customer and Know your Provider sets of rules. Kramer
is also facing a fine could be up to around
six million dollars. These penalties are meant to send a
message to folks who are considering a similar scheme that
if you do this kind of thing, it's going to
(15:07):
cost you. As for Lingo, the company also agreed to
make some changes to how it operates to WITT, weeding
out spoofed phone numbers and only presenting a number when
Lingo can verify that it's exactly where a call is
really coming from. Presumably you would otherwise see something like
unknown caller or something like that on your caller ID.
(15:28):
Lingo must also verify the identities of customers and work
with upstream providers that have quote unquote robust robocall mitigation.
So yeah, this is really sitting a message of saying
this kind of thing will not be tolerated. In twenty
twenty three, GM's cruise business shut down effectively after one
of its autonomous robotaxis dragged the pedestrian for twenty feet
(15:51):
in San Francisco before it came to a stop. The
pedestrian had already been struck by another car that one
was operated by a human. Pretty awful, like a really
hard ruble sequence of events, and CRUZ faced a massive investigation.
The CEO of the division promptly jumped ship, as did
several other leaders, and GM laid off a significant number
(16:11):
of workers within the Cruz division. But now Cruz is
back in the news, having struck a partnership deal with Uber.
New CEO Mark Witten said, quote, we are excited to
partner with Uber to bring the benefits of safe, reliable
autonomous driving to even more people, unlocking a new era
of urban mobility end quote. Which is interesting to me.
(16:34):
Uber itself really was aggressively pursuing robotaxi strategies several years ago,
because I mean, cutting human drivers out of the equation
means more money for the home office. Am I right?
But snarky comments aside. Uber pretty much pulled the plug
on its own efforts after a tragic incident in twenty
eighteen involving another pedestrian accident. Uber then switched to partnering
(16:57):
with companies in the autonomous vehicle space rather than pursuing
their own program. So can these two companies, each with
blemishes on their respective records, team up to create something
that's safe and reliable. Cruz is currently conducting autonomous vehicle
testing with supervising safety drivers in cities like Houston and Phoenix.
No word yet on when those driver lest Uber rides
(17:19):
will become a reality, or specifically which markets that might
happen in I know what you're thinking, you know, Jonathan,
it's been a hot minute since I've learned about a
new streaming video service launching. Well, rumor has it that
we might be getting yet another one from a seemingly
unlikely source. That source is Chick fil A, the fast
food restaurant known for chicken, among other things. Deadlines Peter
(17:41):
White reports that Chick fil A is planning a service
centered primarily around reality television and unscripted content and game
show programming, all with like a family friendly focus. Presumably
the programming on this service would in some way advertise
or promote the company, perhaps through the production of branded content.
How many folks are out there itching to sign up
to yet another streaming service, let alone one spearheaded by
(18:04):
a restaurant company. I have no clue. I'm not going
to write it off just yet, as it could always
surprise me. But my first impression is this is going
to be a very tough sell, particularly during a time
when people are already taking a harder look at their
family budgets for stuff like entertainment. Jesse Kiff is in
hot water for hacking into a government registry in Hawaii
for what purpose? Faking his own death. Kiff hacked into
(18:26):
the Hawaii death registry system and marked himself down as
previously alive or no longer breathing, or debt as a doornail.
To be honest, I don't know what the checkboxes actually say,
but the point is Kiff was faking his own debt.
He was also using fake credentials in an effort to
secure a credit card or debit a card account. It's
awfully hard to navigate the modern world unless you've got
(18:47):
access to that cheddar. So why was Kiff doing all this? Well,
apparently it was in order to avoid having to make
child support payments. He's already been tried and found guilty.
He faces a prison term of sixty nine months. Nice
he'll have to serve eighty five percent of that sentence,
after which he will be released, but will remain under
supervision for three years. Now for some recommended reading, So
(19:08):
first a recommend checking out Eric Berger's piece for Ours
Tetnica titled Against All Odds and Asteroid Mining Company appears
to be making headway, which is a cool story. We've
been hearing about potentials for asteroid mining for several years now,
so it's neat getting an update. Next up, Patrick George
has a piece in the Atlantic titled The Hardest Sell
in American Car Culture, and it's about how the Ford
(19:29):
Car Company wants to encourage American car shoppers to think
about smaller vehicles rather than the trucks and SUVs the
industry has kind of migrated to in the US over
the previous years. And that's because smaller cars are lighter.
Lighter cars are easier to move, and that means the
battery requirements for EV's that are smaller are more manageable
than for those big old chonkers that are currently favored
(19:51):
by the US. So if the US is to move
to more evs, part of the picture may also mean
driving smaller vehicles. That's it for this week. I hope
you are all well and I'll talk to you again
really soon. Tech Stuff is an iHeartRadio production. For more
(20:13):
podcasts from iHeartRadio, visit the iHeartRadio app, Apple Podcasts, or
wherever you listen to your favorite shows.
TechStuff News
Host
Jonathan Strickland
Popular Podcasts
1. 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.
2. Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations.
3. Crime Junkie
If you can never get enough true crime... Congratulations, you’ve found your people.