Artificial Intelligence

[Pete]

Do you think something will pass the the Turing test (30% of human testers think it’s human) consistently in our lifetime?  Or put differently, do you think I’m our lifetime we will have social issues about the rights of artificial life?

[BIG APPLE CAT]

My biggest concern about AI is the potential for like, literally nothing to be reliable evidence of anything. Like any photo, audio, video, video with audio…all of it could seem so real but it’s just uncanny deep fakes. The propaganda potential is terrifying. And like procedurally in a court of law…seems like just about anything other than DNA evidence could be argued  that it was AI generated bc there would be no means of authentication. To me this is equally as scary from the standpoint of opportunistic defense attorneys trying to create reasonable doubt, but also innocent people being framed bc of really good AI

[DaBigTrain]

We already have stuff passing the turing test. It's really not a good measure though to determine if something is really good. There are so many other factors in it. I think tho that we are on the cusp of something. ChatGPT was like a bomb going off and we will probably see an absolute explosion in it during the next couple of years.

[sys]

do you think I’m our lifetime we will have social issues about the rights of artificial life?

no.

[BIG APPLE CAT]

I just hope they abide by the first law of robots which is: their eyes have to turn red before they do anything evil

[Spracne]

My biggest concern about AI is the potential for like, literally nothing to be reliable evidence of anything. Like any photo, audio, video, video with audio…all of it could seem so real but it’s just uncanny deep fakes. The propaganda potential is terrifying. And like procedurally in a court of law…seems like just about anything other than DNA evidence could be argued  that it was AI generated bc there would be no means of authentication. To me this is equally as scary from the standpoint of opportunistic defense attorneys trying to create reasonable doubt, but also innocent people being framed bc of really good AI

Some decent posits. I wonder if this will accelerate the use of blockchain mechanisms for authentication.

I've been using AI tools for years to help with legal work. ChatGPT is on a different level though, and it's a tool that has the potential to save a LOT of time, in terms of coming up with the bones of various documents. I don't believe it's close to replacing human reasoning, because we humans are fundamentally irrational. But I could easily see it generating efficiencies that obviate the need for some X number of jobs. I'm trying to be at the vanguard of utilizing it so as not to be left behind.

[Pete]

Will we ever be able to upload our consciousness into a man made machine? Would that come before or after man made artificial sentient life?

[BIG APPLE CAT]

My biggest concern about AI is the potential for like, literally nothing to be reliable evidence of anything. Like any photo, audio, video, video with audio…all of it could seem so real but it’s just uncanny deep fakes. The propaganda potential is terrifying. And like procedurally in a court of law…seems like just about anything other than DNA evidence could be argued  that it was AI generated bc there would be no means of authentication. To me this is equally as scary from the standpoint of opportunistic defense attorneys trying to create reasonable doubt, but also innocent people being framed bc of really good AI

Some decent posits. I wonder if this will accelerate the use of blockchain mechanisms for authentication.

I've been using AI tools for years to help with legal work. ChatGPT is on a different level though, and it's a tool that has the potential to save a LOT of time, in terms of coming up with the bones of various documents. I don't believe it's close to replacing human reasoning, because we humans are fundamentally irrational. But I could easily see it generating efficiencies that obviate the need for some X number of jobs. I'm trying to be at the vanguard of utilizing it so as not to be left behind.

Funny you say that, as I was typing that out I was thinking they’re going to need some kind of ledger like block chain authentication that can’t be removed. Crazy to think of a video needing the same kind of provenance as a Picasso.

Also it seems like it could go a long way in terms of taking a first pass at writing briefs, provided it gets heavily proof read and edited. But I like the idea of my counsel having more time freed up to muster my vigorous defense than the tedium of brief writing and other such minutia

[catastrophe]

My biggest concern about AI is the potential for like, literally nothing to be reliable evidence of anything. Like any photo, audio, video, video with audio…all of it could seem so real but it’s just uncanny deep fakes. The propaganda potential is terrifying. And like procedurally in a court of law…seems like just about anything other than DNA evidence could be argued  that it was AI generated bc there would be no means of authentication. To me this is equally as scary from the standpoint of opportunistic defense attorneys trying to create reasonable doubt, but also innocent people being framed bc of really good AI

Currently all these things have to be authenticated by sworn testimony from someone who can attest that they’re accurate representations. Doesn’t mean forgery/fraud won’t be more common, but people will always have to stick their necks on the line for it (unless/until we’re satisfied that some new technology like blockchain does the job).

[8manpick]

We’re all mumped. The only question is the timeline.

[Pete]

We’re all mumped. The only question is the timeline.

There is still a chance it will think we make great pets.

[BIG APPLE CAT]

My biggest concern about AI is the potential for like, literally nothing to be reliable evidence of anything. Like any photo, audio, video, video with audio…all of it could seem so real but it’s just uncanny deep fakes. The propaganda potential is terrifying. And like procedurally in a court of law…seems like just about anything other than DNA evidence could be argued  that it was AI generated bc there would be no means of authentication. To me this is equally as scary from the standpoint of opportunistic defense attorneys trying to create reasonable doubt, but also innocent people being framed bc of really good AI

Currently all these things have to be authenticated by sworn testimony from someone who can attest that they’re accurate representations. Doesn’t mean forgery/fraud won’t be more common, but people will always have to stick their necks on the line for it (unless/until we’re satisfied that some new technology like blockchain does the job).

If only perjury was more of a deterrent.

[cfbandyman]

I don't love it, but I also don't know how you "stop" it. I mean, development will happen by people no matter what happens, I just hope we know what we're doing (not a good feeling about it)

[BIG APPLE CAT]

one thing i don't think AI will be able to do: high fidelity forgery of hand writing.

Source: i have a Cricut and to see how that thing "writes" is wild.

[schreds21]

Apparently BAC is the only other person in this thread besides me who has seen any of the Terminator movies.

[Rage Against the McKee]

That's a good point. The terminator was famously bad at forging signatures.

[BIG APPLE CAT]

That's a good point. The terminator was famously bad at forging signatures.

In the original script the terminator was supposed to be able to forge handwriting perfectly but Arnold is like, notoriously bad at forging handwriting so they had to re-write the script midflight b/c they had too much of the film already "in the can" to justify re-shooting with Lou Ferrigno as the Terminator (very good at forging handwriting).

[MadCat]

Mankind is just one iteration of sentient being creating a self-perpetuating artificially intelligent being which continues the cycle throughout time 

[Pete]

That's a good point. The terminator was famously bad at forging signatures.

In the original script the terminator was supposed to be able to forge handwriting perfectly but Arnold is like, notoriously bad at forging handwriting so they had to re-write the script midflight b/c they had too much of the film already "in the can" to justify re-shooting with Lou Ferrigno as the Terminator (very good at forging handwriting).

Good work here, everyone.

[Pete]

Mankind is just one iteration of sentient being creating a self-perpetuating artificially intelligent being which continues the cycle throughout time 

OR ARE WE?!?!?

[passranch]

We’re all mumped. The only question is the timeline.

There is still a chance it will think we make great pets.

I'm just here to appreciate the elite Perry Farrell reference.  Nicely done.

[Pete]

We’re all mumped. The only question is the timeline.

There is still a chance it will think we make great pets.

I'm just here to appreciate the elite Perry Farrell reference.  Nicely done.

It means a lot that you noticed!

[Dugout DickStone]

We’re all mumped. The only question is the timeline.

There is still a chance it will think we make great pets.

I'm just here to appreciate the elite Perry Farrell reference.  Nicely done.

Was that what they were talking about?

[Pete]

We’re all mumped. The only question is the timeline.

There is still a chance it will think we make great pets.

I'm just here to appreciate the elite Perry Farrell reference.  Nicely done.

Was that what they were talking about?

Aliens, but AI works too

[Pete]

I’ve been listening to Lex Fridman podcasts about AI, and I have a couple more to listen to in order to round out the view points on this, but at this moment I think the best case scenario is the following:

1. Everyone is able to gain access to a personal AI (e.g one one your phone/earbuds/augmented reality glasses/etc), and there is no centralized single AI for everyone.

2. The average IQ of humans is boosted 20 - 40 points when they have access to an AI.

3. Violent crime is significantly lower in populations of higher IQ, and we realize those benefits

4. Jobs are not taken by AI.  Jobs are taken by people with AI.

5.  Terrorists with AI are stopped by good guys with with AI.



The worst case scenario is probably:

1. Centralized AI controlled by a government party. Social scoring and civil compliance is monitored by the AI.  This is bad if you are someone who deviates from the average, such as LGBTQ or a different religion from the sanctioned version.  If you are in alignment with the values of the organization who trains and aligns the AI, you could be quite content, however.

2.  Rouge States and bad actors use AI to engineer bio weapons and deploy them against rivals.  They also use AI to break encryption to steal wealth and disrupt markets.

3.  While AI may result in job growth in the long run, many could be the victims of transient unemployment as jobs are eliminated in the “laptop” class and they need to retrain or find lower wage work as laborers.

4. Possible elimination/massive reduction of the middle class, but also possible lifting of the lowest class.