Economic Update with Professor Richard Wolff
99.5 FM and WBAI.org online and it's now 7 p.m.
Stay tuned for Off The Hook coming up.
Phone ringing.
Beep.
Thank you for listening to Off The Hook.
The telephone keeps ringing, so I ripped it off the wall.
I cut myself while shaving, now I gotta make a call.
We couldn't get much worse, but if they could they would.
One million pound for the best, expect the worst.
I hope that's understood, one million pound.
Beep.
Thank you right there and I'm Skype land we have Rob T firefly
Good evening right here, and we have Gila good evening right over there, and we have Alex
Good evening. Oh good. I'm glad we're all here. It's uh. It's nice on this chilly night
Boy, we're not on next week so everybody make note you can you know take a vacation or whatever
back in two weeks
We do want to encourage people to support the radio station which keeps our voices flowing and you can do that very easily by
Going to give to WBA I org give the number to WBA
I org please take this request seriously because a lot of us don't and a lot of us
Just assume that things will survive forever, and they won't they won't without your support
So if you're one of the people listening right now
That said yeah, it doesn't really apply to me because the station will be fine
Don't assume that don't assume that give to WBA
I org is the website on that website you can find all kinds of amazing premiums and
Things that you can get for your pledge and your pledge can be a really really tiny amount
That's a monthly payment which adds up to a bigger number. That's how addition and months work
so 12 times
10 that's 120 that's great, and that's something that if we got a hundred people to do that
It's a significant amount and it keeps the radio station on the air and somehow
This radio station has been on the air since around 1960
Non-commercially
presenting all kinds of alternatives, and you know especially in these times boy there are a lot of alternatives out there a lot of
Different ways of looking at the world well
This is the place to go to find it and to find people arguing back and forth constantly about everything
Yeah, that's that's what it's all about, but we survive and we survive through your help so once again give to WBA
I dot o RG
All right that said
Has anyone had any fun times this this way I've been I've been
Getting more familiar with different forms of social media
I have to say I'm more impressed with blue sky over the past week because they've been responsive
I was complaining. They weren't responsive and now they are responsive and
Various issues that we had they have dealt with which is great
The only thing you know I've got these
These invite codes nobody seems to want them, and I don't know yeah
I can't convince you guys to take them so you know that doesn't bode well
We're all on Twitter a lot of us are unmasked it on I suggest giving blue sky a look because it has it has some
Advantages it seems to be a lot easier to use. I don't know can I can I convince anybody here?
Well when I said it, I didn't mean ever
You know I didn't want it that minute you stomped your foot and that kind of implied forever, so I don't know
How about Alex what about you? I think it's a hard pass
Hard pass for me as I'll reiterate my
Thoughts on social media being a useless cesspit of noise with little to no redeeming signal once again
But it's our cesspool of noise with no redeeming. I mean
We have to we have to have home turf someplace to Rob Gila
Well, you know when when you offered those invite codes. I I didn't say no. I I made a fart noise, but I
Know I I I am perhaps a little more mildly curious now that I hear that you're having such a good time over there
although I think my my heart and soul will always
veer toward the open-source federated nature of mastodon over any closed service for the
Foreseeable well, I'm using that too, and I'm having fun over there as well
Haven't quite figured out though how to get people to migrate
Because you know where where I had a lot of
followers before
very few now and
Not sure how many times I can say on on the other outlets. This is where I am on this particular one
Or do I have to call people? How do you how do you get people to migrate over and and to continue the conversation?
That's that's a difficult part and of course
2600 has a presence on all of these now
And it's a little difficult because you know we have to like you know say for the meetings coming up this Friday
There are hacker meetings all around the world first Friday of the month
We'd like to announce that on Twitter well now we announced it on Twitter
And we announced it on mastodon, and we announced it on blue sky. There's a lot of announcing going on
We need a full-time announcer basically that was the the lion's share of my reaction. It's just more social media
Elicited that response right it is a lot of work creating the account and then maintaining
with some sort of
Content and all that it just it just seems like it's never ending this
Facebook and Instagram and whatsapp and what's not and all it just where does it end where does it stop?
There are a lot that you know I refuse to even get involved with I'm with Alex on that
But you know there there has to be some way to reach out to people besides
The radio which we're fortunate enough to have most people aren't
So how do you it used to be easy used to be so easy everyone had?
Instant messaging and you could just talk you know
Without worrying about what platform somebody was on Gila
Have you considered renting a plane with a banner you see it's it's it's suggestions like that that make me realize that
just barking into the wind here and
We're all doing
Which is is there some kind of
Content creation platform that you can go in one place, and then it will sort of
Like a champagne fountain you can pour it into one place, then it will spigot out your stuff into all its various cups
So you can post it once and then it will say hey take this to Mastodon to Twitter to blue sky to
Facebook to Instagram to Twitter to I don't know the bottom of somebody's shoe with chalk
Whatever it may be do those places exist well
I know they used to exist
I'm sure there is something like that that exists now
And I'd love to hear about it, and please if you know about it write to us OTH at 2600 comm in the past
There used to be a really simple way to have your tweets go to your Facebook account
And that was the only interaction I had with Facebook because I hate Facebook
but I would I would send a tweet and that tweet would show up on Facebook and
People would see it there, and they might reply to it, but I wouldn't see that because I didn't log into Facebook
But then Facebook for some reason clamped down and stopped allowing that although what they're doing now is they're sending me memories
Every day a memory of something that I said
Ten years ago and many times those quote-unquote memories are our tweets that were
Reproduced on to Facebook, so they're kind of pointing out
How much cooler they used to be in fact one of them was me saying hey, it's my 2600th tweet on Facebook
Yeah, so that's a big waste of time. I know for a lot of people, but
it is important to communicate and
These days there are a lot of people who don't even use email anymore. How do you communicate? We don't make phone calls
We don't send emails. We're all on different platforms. We're splintered into a thousand pieces
you know
Short of gathering in person, which is you know why we have the meetings and the conferences
If you know we need to develop better practices, I think
Rob go ahead
It was interesting because I recently had to
Design myself a new business card it was time to do that so I was putting this together for my order and
Realized that you know in the old days you would put your Twitter
Username your your whatever social media on your business card so people who are interested in that could find you
I couldn't think of anything that I really wanted to put on there except just my own
Personal website which has links to whatever else I may be using at the time
But I think that's that's the only evergreen way to do it now because you don't know what other services are going to be there
tomorrow
Alex good
You know to give us point about
Consideration of renting a plane for something I want to remind you in particular manual that I think you actually did do
I did do that. I regretted it. This is for the free Kevin
rallies that we were having and
We really we let our heads get really big and and think that yeah, we can have the whole city see this
and
We hired a guy with a skywriting plane to to say free Kevin in the sky. It was a cloudy day
Nobody saw it
We still have to pay the guy
And have not done that since yeah, I mean it's the thought that counts that coincided with the free Kevin protests that
Occurred outside the federal courthouse in southern Manhattan and was I think a huge success?
But another thing in terms of reaching people have you considered maybe thinking about this old school maybe putting out a magazine or something?
Well, you know that is a possibility
Although I understand that print is dying as well. Everything is dying print is dying
Radio is dying all the cool things. We're being told are dying, but it's not true
it's it's true of maybe the
the unimaginative and
The carbon copies, but there is always going to be a place for people to express themselves
in in creative ways over
Traditional broadcasting or printing methods and I dare say online as well. Go ahead Rob
You know walking around my neighborhood and running errands today
I did notice that people are still signing graffiti with their Instagram handle, which is interesting
I would have thought that's a good way to get caught for doing graffiti, but I don't know how these things
What do you know? It's their Instagram handle not their Twitter handle because they look exactly the same
Well, at least one of them has Instagram colon and their their username
So, okay, if they indicate it then that makes it a lot easier, but you know that brings up another issue
we've been trying to to spread the word about our
2600 blue sky account now blue sky is cool because
You if if you have a domain and we do 2600 calm you can make your domain your username
Otherwise, it's it's username dot. I think it's BK sy
Dot social. I think that's that's what the blue sky address is
What we cannot seem to find out is how do we share the blue sky address?
With people when the domain is the username. So how do we say?
We're 2600 calm
Like what's our address on blue sky? How do we do that? Because I don't think it's 2600 calm dot BKS. Why not social?
I don't think that works
so nobody seems to know the answer to that so
It's it's still in the infancy. I think figuring out how it all works
There's when you click on your own profile is does it show you an interesting URL it just shows 2600 calm
So that doesn't tell us much and and we can't just tell people that's our address on blue sky because obviously that's that's our website
And it's a little confusing
so we just need a way of being able to give people an address and say that's blue sky and that's
That's Mastodon and that's Twitter and that's whatever else. So
All right, moving on. We have a lot of news to get to today
Interesting story that was posted to our list
Concerning something called nightshade, which is kind of a response to chat GPT
since
since that
Curse, I guess burst onto the scene about a year ago
It's been about a year since chat GPT made its debut the generative AI era has kicked into high gear
But so too has the opposition and that's what we're talking about here
A number of artists entertainers performers and even record labels have filed lawsuits against AI companies some against chat GPT maker open AI
Based on the secret sauce behind all these new tools training data
That is these AI models would not work without accessing large amounts of multimedia and learning from it
including written material like the 226 hundred books that were accessed without knowledge without permission of this and
Images produced by artists who had no prior knowledge nor were given any chance to oppose their work being used to train new commercial AI
Products. So this article in venture beat comm
Goes on to say in the case of these AI model training data sets
Many include material scraped from the web, which is a practice that artists previously
By and large supported when it was used to index their material for search results
Which now many have come out against because it allows the creation of competing work through AI
Now without filing lawsuits artists have a chance to fight back against AI using tech
MIT technology review got a look at a new open source tool
Still in development called nightshade, which can be added by artists to the imagery before they upload it to the web
altering pixels in a way invisible to the human eye
But that's poison for the art that any AI models are seeking to create
Yeah
nightshade was created by
University of Chicago researchers and the computer science professor Ben Zao and
Will be added as an optional setting to their prior product glaze another online tool that can cloak
Digital artwork and alter its pixels to confuse AI models about its style
In the case of nightshade the counter-attack for artists against AI goes a bit further
It causes AI models to learn the wrong names of the objects and scenery
They are looking at for example
The researchers poisoned images of dogs to include information in the pixels that made it appear to an AI model as a cat. Oh
I love this after sampling and learning from just 50 poisoned image samples
The AI began generating images of dogs with strange legs and unsettling appearances
after
Wow after a hundred poison samples, it reliably generated a cat when asked by a user for a dog
After 300 any request for a dog returned a near-perfect looking cat
You know
This is this reminds me of when I was a kid and I wanted to create a calculator that would only give the wrong answers
You know and no one else seemed interested in that but that would have been really a cool device
And if anybody has that please also let us know
The researchers use stable diffusion an open-source text-to-image generation model to test nightshade and obtain the aforementioned results
Well, you can see the potential here and I think this is just great I think this is you know kind of related to
Answering with with bad information when given security questions for instance, you know, where did I grow up if you say Neptune?
Nobody's gonna be able to you know, figure that out. That's of course a private question, but whenever you are asked
any kind of personal information by
Anyone you don't know anyone who's a stranger or a computer you you lie and by lying
We have we have we have an Alexa device in the next room whose name is computer
And and he just started talking to me when I when I made that suggestion
So yeah, we have to keep our voices down because this could get kind of scary but there there this is the plan to fight
back I
Love this so much. I love how this is basically jamming the radar and
Having you know, it's it's a bit of mischief and it's a bit of mischief. That's uh to the end of doing something useful
I think for artists like myself who are dubious about the value of just letting AI index your stuff
Willy nilly and I had been hoping for something like this and it happened and I'm so I'm so happy
I want to play with this. I
think this builds a little bit of integrity into it and
maybe sets up a better picture of letting people understand and have confidence in the kinds of things that are used to build the
build these
Algorithms and kits or learning models they they
And
Instead of the tendency to just throw anything and everything into it
you could have some formality or some process or at least permission and then the
appropriate art or whatever content that is not a filtered in this way could be
Contributed could contribute to the overall project under
Those kinds of
Guidelines and and then that could be a part of people understanding the the the model itself is is when they can trust
Alex if I could fire up your legal mind for a moment. Do you foresee possibly?
the industry
fighting back against this saying that this is some kind of vandalism that that you're not permitted to
Pollute the data stream and and lie to AI. Do you think that there's actually a legal case there?
Yeah, you took the thoughts right out of my head there in that I do think there is an argument to be made that
by poisoning the data knowingly inserting poison data or malicious code in some respects into
The AI model you could be in in essence violating computer tampering laws. These are laws that are mostly
at the state level
Rather than the federal level. It's not necessarily the computer fraud and abuse act which requires unauthorized access to something
We're breaching some kind of security protocol in order to get over to going through a gate or something like that or to access data
so this is different from that, but this is more of like a
tampering kind of
situation where I could see that the
AI companies would
Want to try to enforce their rights that way the problem that they're going to have them
Is that it's not up to them to enforce the criminal laws. It's up to the government and
The AI companies have not necessarily been the best bedfellows when it comes to government regulation
In fact, they've been snubbing their nose at governments, especially the EU
in terms of regulation
And you also have the Biden administration just two days ago issuing the executive order with respect to use of artificial intelligence
Guidelines cybersecurity health care, etc. So there's a lot of that coming down the pipeline
I wouldn't I wouldn't hold my breath if I were the AI AI
Industry looking to enforce the criminal laws in this respect. What I'm also interested in is
What kind of defenses they are going to then employ to counteract this type of poisoning?
Are they going to include some kind of additional screening additional tests and exactly how does this attack work?
Isn't the injection of some kind of metadata into the file itself?
that that would certainly
Interest me. I'd love to know more about how this works. All right, Rob
Yeah
I have trouble with the idea that this could be any sort of criminal act just because
What it's doing if I'm reading the article, right?
Is it's just making subtle changes to the actual pixels of the image the image data, which that which don't make a difference to the human
eye, but which computers notice and AI algorithms scraping your image would notice and
through that it's it's manipulating what the AI interprets the image as and
I think I am as someone who puts imagery online
I think I'm allowed to put whatever funky pixels
I want in my in my artwork without having to worry about whether it's a criminal act because somebody I never signed any sort of
Agreement with or agreed to any terms of service with doesn't like what it does to their their chomping machine
I think I think the you know, I am NOT the lawyer in this room, but I have trouble not believing that this is quite
Quite a quite a simple non-criminal manner
Well, it's just interesting that good old human lying might be what it takes to keep AI from becoming
Superior over all of us or the dominant word or authority figure
I mean, it's already doing a good enough job as it is of getting things completely and spectacularly wrong
I mean making up facts making up people making up all sorts of history
It's doing that all on its own. But if if we tried to derail it even further
It would become rather useless and and well useful for some
Performance art I suppose but not really useful for
Well, you wouldn't have lawyers showing up in court citing cases that didn't exist anymore
So I think maybe people would stop trusting it. Yes, Alex
Yeah, I tend I tend to agree with Rob here
I think that the the case of this would cross the line and criminal law is somewhat
It would cross the line in terms of criminal law somewhat tenuous
What what I do think is is really interesting though
Here are our artists taking into their own hands the right to protect their their images in some respect
now however
Hopefully this would protect that artist's actual work from being ingested in a manner by that AI
That could allow the AI could to then replicate things in the style of this particular artist
Let's you know call it Picasso something if Picasso were here anyone protect all of his images
He would alter the pixels as Rob indicated with the nightshade program
They then get ingested into something like stability diffusion and then stability diffusion presumably would have difficulty
Generating images in the style of Picasso if that's the case
Then I think this could be entirely kosher here because you're just protecting your own artwork
if if however it
Bled into
affecting the AI in general and if the AI is ability to generate let's say images of
starry nights or images of
Cats or dogs or something like that?
Then I think there is more of a case when it comes to computer tampering laws
And maybe even crossing into the the crossing threshold of malware
so
You know, I think it remains to be seen how this works on the other hand. I think it's unlikely
This is I guess the third hand now
But I think it's rather unlikely that any individual artist who uses nightshade to protect his or her own images
Would have some kind of overall effect on the artificial intelligence engines ability itself to generate new images
Gen generate new images generally and I think that's because
They're trained on billions and billions of images as opposed to just singular
Yeah, but um if if artists on must start adapting start adopting use of a tool like glaze or whatever comes up because I
Think what we're seeing is the beginning of what's going to be like an arms race
Between people who want to mess with AI and the AI companies who wanted to work the way they wanted to work
But I think like from an artistic perspective. I mean the Belgian surrealist painter Rene Magritte in
1929 painted the very famous painting of a picture of a smoking pipe
With the text underneath indicating this is not a pipe
And I think if I put up a picture of a dog and make a computer think it's a cat
That is it's not my fault that the computer can't artistically interpret the work in the in the spirit in which is it is intended
You know we come back to Orwell's famous 2 plus 2 equals 5
Well, sometimes it does when when Big Brother tells you that 2 plus 2 equals 5 then 2 plus 2 equals 5
There are people in this world who will believe that who will you know?
Follow what they are told to believe there are people
In government who believe I'm not gonna get any further than this but
This is basically twisting the truth to suit your purposes
We all do it in one form or another now in fairness. We reached out to Google Bart on this issue
asking if if
Telling lies is something that it's concerned with
And it responded I believe that it is important to be honest and truthful even if it is difficult
Lying can have negative consequences for both the person who is lying and the person who is being lied to
It is better to be honest and face the consequences and to lie and risk damaging trust and relationships. Yeah, I kind of think
Bard didn't really
Comprehend what we were asking because it's not about
People's relationships, it's about us lying to you Bard and and and and screwing up your
your data and and the way that you process facts and
Maybe Bard is still naive enough to believe that people will always
Tell them the truth
Who knows?
This is the right I think kind of tool and many others like this. I think will become
popular or at least intriguing as this technology is is
Rolled out people will want these kinds of things because we've talked so much about
Okay, my art was already online. It was already posted
It it's in the past and was used now going forward. I'm finding myself
I'm sure many other people are thinking about this as they're moving platforms or on different sorts of social media sites that we're
What images am I publishing going forward and how am I preparing them?
Because I know that if they're public one way or another they may end up in
Some
private or public
model like this
Whether you know it or not and and act accordingly
So these types of things maybe are ways another step in how you might prepare an artwork
for publish for public dissemination and publishing online, but
It's very interesting the the types of things that may may become necessary may become like
Completely normal in the future. I just asked Bard and and and this had it spinning for about 45 seconds
What if 1,000 people tell you the same false fact?
I've never seen it do this before it gave me two answers draft a and draft B and a little
Listen icon, so we might actually be able to hear Bard answer. Let's see if this works
Now if 1,000 people tell me the same false fact
I would be more likely to believe it, but I would still try to fact-check the information before accepting it as true
I would also consider the following. Why are so many people believing this false fact?
Is it because it is being spread by a credible source? Is it because it is emotionally appealing?
Is it because it confirms people's existing beliefs? What evidence is there to support the false fact?
Is there any scientific evidence? Is there any eyewitness testimony? Is there any logical reasoning?
What evidence is there to refute the false fact? Is there any scientific evidence? Is there any eyewitness testimony?
Is there any logical reasoning?
If I'm unable to find any evidence to support the false fact or if I find evidence to refute it
Then I will conclude that the false fact is false. Even if 1,000 people believe it
It is important to remember that just because a lot of people believe something does not mean that it is true
False information can spread quickly and easily especially online
It is important to be critical of the information that you consume and to be willing to question things even if they are popular beliefs
Here is an example of how I might respond to a situation where 1,000 people tell me the same false fact
1,000 people the earth is flat
Me, I'm not sure. I believe that I've seen pictures of the earth from space that show that it's round
1,000 people but those pictures are fake. Me, why would they be fake?
1,000 people because the government is trying to hide the truth from us
Me, that doesn't make sense. Why would the government want to hide the truth about the shape of the earth?
1,000 people. I don't know but it's true. In this situation
I would be skeptical of the claim that the earth is flat even though 1,000 people believe it
I would try to find evidence to support the claim, but I would also be open to the possibility that it is false
If I'm unable to find any evidence to support the claim or if I find evidence to refute it
Then I will conclude that the claim is false
Wow, I was not expecting that that was very in-depth and there's a second answer to which we don't have time to play on
Overtime we'll play it. Alright, so tune into overtime at 8 o'clock on YouTube and you'll hear Google Bards second response
But that I was kind of impressed by that
Alex, what do you think? I
Thought it was actually very cogent. I thought it was well thought out and
Thought about assumptions biases predisposed notion longest answer I ever got from that thing
examples relevant examples
Numbers it was actually quite a good response Aguila. Did you know I did because
It's interesting to me to see where stuff is coming from because when I heard this
Explanation the first thing it made me think of was
Lady Gaga in the run-up to the Oscars in 2018
because
people were asking her about this movie and she kept saying there can be a hundred people in the room and
99 people don't believe in you but one does and
You have to and like one person believed in me and that was Bradley Cooper every single interview
She gave for like the better part of a year. She said exactly that so that was the first thing that I thought of
When I heard Bard's response, I don't know what that says about me
I'm not really excited to contemplate that but it I didn't know that they trained Bard on Lady Gaga's interviews. That's all I'm saying
It's quite possible. That's exactly what happened. But you know, I think Bard might be a little naive here because
It isn't always so black and white as as to what is true and what is not true
Even if a thousand people are saying something else, it's more nuanced, you know, we see it every day in the news
And you know, we've had the conversations where we try to convince somebody that
No, that's not true because of this and they never quite believe you and you never quite convince anybody
So, I don't know. It's gonna be interesting seeing what's ahead now
We also have this interesting story from wired as the generative artificial intelligence gold rush intensifies
Concerns about the data used to train machine learning tools have grown
Artists and writers are fighting for a say in how AI companies use their work
Filing lawsuits and publicly agitating against the way these models scrape the internet and incorporate their art without consent
Now some companies have responded to the pushback with opt-out programs
I give people a chance a choice to remove their work from future models
Open AI for example debuted an opt-out feature with its latest version of the image-to-text generator Dolly
This August when meta began allowing people to submit requests to delete personal data from third parties used to train metas
Generative AI models many artists and journalists interpreted this new process as metas very limited version of an opt-out program
CNBC explicitly referred to the request form as an opt-out tool
But this is a misconception in reality. There is no functional way to opt out of metas
Generative AI training artists who have tried to use metas data deletion request form have learned this the hard way and have been deeply
frustrated with the process
Illustrator Mignon's a Guga says it was horrible
Over a dozen artists shared with wired an identical form letter
They received from meta in response to their queries in it
Meta said it is unable to process the request until the requester submits evidence that their personal information appears in
Responses from metas generative AI
Mihaila Volshu a Romanian digital artist and photographer who has tried to request data deletion twice
Using metas form says the process feels like a bad joke
She's received the unable to process request boilerplate language, too
It's not actually intended to help people
Bethany Berg a Colorado based conceptual artist has received the unable to process request response to numerous attempts to delete her data
I started to feel like it was just a fake PR stunt to make it look like they were actually trying to do something
she says
So as artists are quick to point out
Metas insistence that people provide evidence that its models have trained on their work or other personal data puts them in a bind
Meta has not disclosed the specifics about which data it has trained its models on so this setup
Requires people who want to remove their information to first figure out which prompts might elicit responses that include details about themselves
or their work, so
You kind of have to submit information if you want the information to be deleted catch-22 thoughts
This this also means you have to be using metas, uh, you know using metas a I generation
material, you know software in order to find proof that it's using your work and
So either way, it means you have to go elbow-deep. You can't just like opt out of the whole thing
But I think it's also about what is
Public and what is free and what is global and then what is grabbed?
You know in the same way that I was thinking about fast
TV
Right free ad supported streaming television, which is basically what we used to get over the air
It's TV with ads, but now everybody has an app for it. You have to opt in to it
It's a complete difference. You can't just plug in and go anymore because of the
balkanization and ownership of every single
element of the media
Element of our society and culture now
Nothing is just out there anymore unless it can be monetized and owned in some way
Very true. Yeah
there's there's also the the issue of remixing because like as as an artist myself and someone who puts a lot of creative material out on
Out on the net. I'm very
I'm very active in creative commons work, which is about putting your work out there for free for people to reuse
but an important part of that is the attribution element of yes, I'm remixing this thing here's who made it if you want to
find the original and
You know, hopefully people can reuse my things but also say where they got it
And that's the step that's being skipped by all this AI stuff
I think if there were a way for AI to generate images out of people's artwork, but keep that
Attribution aspect intact there there might be much less or maybe even no fight here against no pushback again
Against this from the artists because you want your work out there and you want your work to be propagated
But you want people to know that it's yours in case they want to buy it from you or hire you or something or at
Least or even just give you a nice pat on the back and say hey good job
Yeah, not much to ask for go ahead Alex. I think this is a really interesting issue here
At first glance it sounds like you know, it's a great idea to allow
People to actually opt out of having their their images scanned and and used to train these types of
systems and data, but one
Once they're in there
How technically feasible is it to actually untrain the data or retrain the data or somehow take it out excise it?
That seems technically dubious to me
But I'm not the kind of guy who's very often training large language models or generative AI
So I'd be curious as to what our listeners think and know about this particular issue
It sounds like Kyle actually might actually have some some insight into this particular. I was just gonna say you aren't yet
Yeah, you're right give me until next week and I'll figure it out now on the other hand though
There have been lawsuits that artists have filed very recently
And in fact this week there was a decision in a lawsuit that was filed
I think it was I know it was against stability diffusion deviantART
I think possibly open AI and I believe it was federal district court in
California and a whole bunch of artists have been together
and you know
full disclosure we helped out one of the defendants in this and the back end not as a council of record, but
Almost the entire suit was thrown out and and the suit was about artists who got together and said hey
We're really angry about our were images having been ingested by this artificial intelligence engine or these engines in fact
And we don't consider that to have been a fair use of our images
We consider it to be also a violation of our right to publicity. It undercuts our ability to sell things that
You know that we make in the future
So, you know if you can just generate art in the style of Rob T
Firefly then doesn't that dilute the market for art for Rob from Rob T. Firefly in some
Respect, but I'll tell you that the the court threw out almost every single one of these clients
But there are defects and all of them except the direct infringement claim. Now the complaint was thrown out almost in its entirety
But not with prejudice meaning that the plaintiffs will have leave to replead that complaint
Replead that complaint cure its defects and refile it and I'm guessing that they will
But this is an issue that I really I can argue it either way
If you're an artist and your image is ingested because it's out there on the internet
You didn't really have an expectation of privacy that nobody was going to ingest this if it's one of billions and billions and billions of images
Shouldn't that be considered some kind of fair use or de minimis use of your particular image on the other hand if you're an artist
Does this tend to undercut your ability to capitalize on your own name and to sell your own works
and if it's ingesting the entirety of the work and it's being used to for the output is being used in some
commercial subscription or for some commercial purpose then maybe this really doesn't constitute fair use of a copyrighted image, so
I really honestly can argue it either way
Wow technology is really is really doing a number on us
We just got this letter from our friend Greg Newby who writes
Hi friends from last week's episode you were talking about Eric Adams robocalls in different languages in his voice
Coincidentally, I've been part of a project Gutenberg collaboration with Microsoft that used the feature of automatic text-to-speech
Using an individual's recorded voice. This was demonstrated at the interspeech conference in Dublin over the summer
You can find the conference paper and project description at aka dot ms slash
Audiobooks in a nutshell. There is a free software stack that leverages Microsoft's synapse ML and other stuff
It can read text ideally after being trained on it
My Microsoft colleagues use this to create audio files of nearly 5,000 books and give them away freely in the demo
Which was also given at the Internet Archives library leaders forum two weeks ago
Is that a person can read a few sentences into a microphone and then their voice can be used to make the
Audiobooks, it only takes a couple of minutes for a full book length audio file
This is now mainstream technology that anyone with some technical skills and perhaps a credit card can access the credit card because
While the software is free not all the Microsoft infrastructure is free to use though. It's in this inexpensive
Possible uses of this technology include creating an audiobook in your voice to read to children while you are away
custom greeting cards
Automatic responders on an office or home voicemail in your own voice
Reading email or other content in the sender's voice
So it's not all about robocalls and deep fakes of famous people and the project
I was part of just one recognition by time as one of the best inventions of
2023 so congrats Greg newbie for that. But boy a lot of things to think about here. I mean, yeah, this is mainstream technology now
It's not all about robocalls and I guess there will be quite a lot of uses
But I don't know. Does that make it much better?
It's gonna be so much fakeness where you think somebody is actually speaking when when they're not Rob
Yeah, I was I was following this whole thing with great interest. I used to work in the field of audiobooks for accessibility
Based libraries for those with say low vision or reading disabilities or things like that that would prevent them from reading a
print book and who depend on audio as an accessible format and
You know at the moment the technology is still at the point where it's out
You can you can pick out that it's not a real human voice. It doesn't stop for breaths
It doesn't you know have the same reading cadence or a meter that a human reader would yeah, but they can program
They can program that in they could and it's it's totally just a matter of time before they do but
It's at this moment
Like these these books that this is generating are
Good for producing a copy of something that is accessible to those who wouldn't be able to access the material otherwise
But it does lack the the element that a human would add to a reading
things like subtle voices for characters and things like that and also
Pronunciations of say advanced technical terms or specialized terms that there's still a lot of mispronunciation
happening in the text, but it's it's a very good starting point and
For and for a library like this, I think it's I think it's an excellent use of the tech
But at the same time I love an audiobook that's written. That's that's read by a person
I listen to a lot of them and I deeply appreciate it as a media format and
It makes me a little sad that
The audiobook narrators of the world will find a little less work for themselves
Yeah, there's that and you know, I guess it's one thing if you're if you're putting your own voice out there, but
What's to stop you from putting anyone's voice out there? It's it's not that hard. You just need a small sample
This is something that is being used in big ways throughout
Microsoft's products
I think we were talking about speaker identification and that is something that through their they're like cloud products is
being integrated into their like
Conferencing with it teams and all these other things. I think we're on one of the things they own this week
I don't know. Maybe next week. They don't own it. You can't keep up with it, but
their novel uses that
They're they're backed up by these these hosting platforms and it's it they're competing with one another for
Who can get the most customers and integrating this stuff Google doing the same thing with Bard into their?
assert software and service platforms as just something that can be accessed in ways that
Teams or whatever conferencing program wouldn't have had
Before but it's sort of augmenting and and kind of boosting it a little bit with extra ability
Well, we're talking about you know, the pleasant uses of this technology
But you know, there are certainly so many nefarious purposes that this can be applied to I mean, you know, you look at
any dictatorship that
you know does not exactly have
free questioning or freedom of the press or anything like that and you have
You know words whether they're confessions by the opposition leader or
Some words by the leader himself
You will never know if it's real or not
how how could you possibly tell and and they are salivating over this kind of
Technology in their hands where they can basically say anything and make anyone else say anything
how do we ever get to the point of knowing when something is true based on all of this unless there are some kind of
Ground rules established early on and I think we might be past the early on stage already Alex
I mean, I'm really reminded a lot of the conversation
We had last week with Welton Chang from from period technologies and we're talking about
Misinformation being promulgated through a lot of these alternative social media platforms
when you combine that with the
Technology that's at our fingertips now where you might be able to take literally a voicemail from somebody and turn that into a business
email compromise attack or
a short snippet of
audio from YouTube and
Make it seem as if somebody is saying something they're not and we're heading into a presidential
Election in the United States and a general election in the United Kingdom all likely within the same year
well half a war raging between Israel and Hamas and Ukraine and Russia and you know
Wouldn't it just be convenient for China to attack Taiwan right now in North Korea to attack South Korea?
I mean the world is already on fire and we are
Trying to fight this battle with several hands tied behind our back here when it comes to regulating social media platforms
I think we really have our hands full. This is going to be a very very big problem for us coming into 2024
well, you know
the the one hope is that everybody has the same access because if one side has access and the other side does not have
Obviously, they all have a tremendous advantage if everybody is able to make you know, whatever other side there is
Say ridiculous things at least it can be proven to the people that what you hear is not necessarily true
It's when you lose that ability. That's when things really start to spiral
You know, maybe maybe maybe that would be true or on the other hand, maybe you just get more and more noise
I think what we need in order to combat this type of technology getting out of control. It's not necessarily more regulation
Well, that might be helpful in some respects
I think what we need are is more technology that would allow us to
Detect when something is generated by an artificial intelligence as opposed to generated by a human being
Some kind of metadata that's injected by generative AI into all these files
Something that we can track and trace and of course there be efforts to remove whatever that that data or pixel
whatever the the hallmark or watermark is of
AI generated data, but I think we need some
Better way of distinguishing between what's real and what's generated by a machine?
But would the watermark be something that was inserted for the AI or for the humans?
You know because if it's the AI anybody can simply
Remove it. It shouldn't be that difficult to do that
If it's if it's you know required that if you're a human making a statement you have to somehow prove you're human
In a way that AI can't I don't know. Would that be effective?
Yeah, I mean that's it I mean, I think it's gonna be a cat-and-mouse game
I think you hit hit it right on the head a manual
but I think if it was an AI that was if it was an AI that was generating some kind of content that could
Be confused with original content. There needs to be some kind of additional data
That's associated with it or associated with a hash of that file so that some third party can verify
Whether or not this is indeed fake or real. I
think we're coming into a
At a point in the world right now where it's it's extraordinarily dangerous to not know what's real and what's fake
So we have to leave this here
But I guess the advice that we have to give to people at this point is don't believe what you hear
Don't believe what you see
It may be completely fabricated and made up because technology is moving so fast and we have no real control right now
but we're all in this together and
We can at least talk to each other in person, maybe we should do that more often
Until robots become indistinguishable from humans. I think we'll be we'll be safe doing that certainly at this week's 2600 meeting Rob
I don't think there'll be too many robots. Most of them will be humans, right?
As far as we know, yes, so, you know that artificial skin and everything is getting very good. We welcome robots
It's not like we're you know, we're telling them to stay away
Well, and we have an abundance of tools and technology so build build systems that that help, you know
Help you that that we're still
Help you meet your your needs and and get get work accomplished
It's it's not just that we have to be
Feeding these sites with content and and things all the time
So build build resilient stuff that it works in ways that that you're happy with. All right
Well, we're gonna have to leave it there our email address OTH at 2600 comm again
Give to WB AI dot o RG give the number to WB AI dot o RG donate what you can keep this place going
We'll see you on YouTube at 8 o'clock
Follow the link on the 2600 comm web page. You can call us you can participate
We'll see you there and we'll be back here in two weeks. Good night
A
Lot of people won't get no justice
A lot of people running and I hide it tonight a
Lot of people won't get no justice
Oh
They both
Oh
Oh
Oh
This is Ralph pointer
Join me and others every Wednesday 8 to 9 p.m. Eastern Time on WBA
I 99.5 on your radio now
It would appear the human movement is such that at any moment in history
There are too few that understand possibilities of existence that would benefit all
Who inhabit this planet and are willing to act on this understanding?
This program will feature that few
What are your views on these issues that impact your life today?
What are your views on America today?
What are your views on America's future?
Can we talk call in to 1 to 2 0 9 2 8 7 7
Wednesday 8 to 9 p.m. On WBA I
99.5 on your radio