I saw this thread on TMP the other day, concerning an AI technology called ChatGPT: http://theminiaturespage.com/boards/msg.mv?id=568651

Basically you can ask it any question you like and it will trawl the internet for information and formulate that into a fairly comprehensive response.  I found it's reply to the Waterloo query in that thread really interesting, especially further down when the questioner asks about specific nations and their impact on the battle.

I don't know whether to be impressed or concerned about this kind of technology.  There was another story this week about a GPT bot that asked the questioner to put some code into his computer so that the bot could attempt to break out of its confines.  I know that the AI obviously isn't sentient and is simply responding as the internet has taught it to, but it still feels weird.

There's also a thread for the Somme here: http://theminiaturespage.com/boards/msg.mv?id=568652 and Zama here: http://theminiaturespage.com/boards/msg.mv?id=568654

I've played with it a bit.
My understanding is that it does not yet have a real time search connection to the internet and relies on data available at a fixed time in the past (2021 I think but could be wrong).

I liked it because it is trainable by making suggestions (and rejecting answers that you know are wrong). By suggesting sources and narrowing focus you can start getting really detailed information not possible with Google. It "remembers" the dialog and you don't have to start over with each question.

Give it universal, real-time access, near infinite cloud storage, and the theoretical processing power of a quantum computer and its game over for humans. Cockroaches and ants will be left to battle it out for the scraps.

There's an interesting article on The Register where, after a few examples of people deliberately 'gaming' the system,  ChatGPT tells the author he is dead, refuses to be corrected, and then invents evidence to back up it's initial statement.

That's rather concerning, I think I'd be a bit confused if a bot told me I was dead and then provided proof of it!  The manipulation possibility by foreign agents is a worry too, that you could feed data into a bot like that to make it spread a false narrative.

Some years ago (c2008 I think) I was looking at the development of AI algorithms for autonomous military drones. There wasn't a lot on open source but there were some interesting papers and I remember one in particular about work being done in MIT. It was quite worrying about what were being thought of as potential developments.

It was definitely heading towards the Cyberdyne end of the market.

The idea was long loiter drones powered by solar panels with self learning AI to identify and destroy targets. That is it didn't have a remote pilot and no-one precisely created an algorithm for the identification of targets. It had an algorithm which allowed it to observe collate and categorise human activities and then build a threat database and act accordingly.

I went back after two or three years to see how many of the glitches had been ironed out (we were at the USMC Boston Dynamics (Hi Google!) LS3 robot dog stage by then I think) and guess what? The paper had disappeared!

What's the relevance to ChatGPT? ChatGPT currently just screws up our view of reality. But somewhere, someone is putting together a military version of Chat GPT or Bing's version which is notorious for sulking and verbally attacking users who disagree with it, with what is happening on the autonomous drone front.

I worry about a kid in Austin flying a drone that attacks something in the Middle East. I'm a lot more concerned about a drone that flies itself and makes its own decisions about what to kill. Especially if it has had a bad day on social media.

[Did someone forget to tell people the Terminator series was a warning and not a description of the way forward?]

I'm not on board with that either, 'sentient' drones that are making their own decisions in life or death situations seems like a recipe for catastrophe. 

I find a lot of this new tech interesting and a lot of it has real-world benefits, but as with anything, it needs to be used correctly.  A loiter drone could be hacked and easily switched to turn on its operators and then you've got a whole new world of friendly fire that can't be traced back to the hackers.

My kids think I'm odd when I talk about a lot of this stuff but I've had a theory for a while now about VR and its impact on the world.  I can easily see the next generation becoming so au fait with VR that the virtual world becomes the norm.  If you can look however you want, live wherever you want, drive whatever you want in VR then why stay in the real world where you're out of shape and can't get a job?  People will start paying VR money to improve their virtual lifestyle and they'll spend more and more time in it.

We'll eventually reach a point where you can pay a service to come to your real house and refill your sustenance feed and empty your waste bag for you, while you're away in a fantasy world 24/7.

Quote from: Leon on 20 March 2023, 05:10:13 PMMy kids think I'm odd when I talk about a lot of this stuff but I've had a theory for a while now about VR and its impact on the world.  I can easily see the next generation becoming so au fait with VR that the virtual world becomes the norm.  If you can look however you want, live wherever you want, drive whatever you want in VR then why stay in the real world where you're out of shape and can't get a job?  People will start paying VR money to improve their virtual lifestyle and they'll spend more and more time in it.

We'll eventually reach a point where you can pay a service to come to your real house and refill your sustenance feed and empty your waste bag for you, while you're away in a fantasy world 24/7.

Isn't that the plot of the Matrix 

Quote from: John Cook on 21 March 2023, 09:46:38 AMIsn't that the plot of the Matrix 

ER Yes. On the local news last night a Co-Op in Greater Manchester is using wheeled drones to deliver groceries. Scary

Quote from: Lord Kermit of Birkenhead on 21 March 2023, 10:30:27 AMER Yes. On the local news last night a Co-Op in Greater Manchester is using wheeled drones to deliver groceries. Scary

Indeed.  Watching the Ukrainians use drones to deliver 'things', it gives the threat 'I know where you live' rather more credibility.

QuoteER Yes. On the local news last night a Co-Op in Greater Manchester is using wheeled drones to deliver groceries. Scary

Ha, i bet the local 'erberts are going to have great fun with that!

A lot of the cheap downloadable books on Amazon are apparently ChatGTP output.  It takes about 30 minutes to produce a novel.  There's calls for it to be made clear whether the book being sold is written by a human.

Quote from: John Cook on 21 March 2023, 09:46:38 AMIsn't that the plot of the Matrix 

Pretty much and we know how that ended!  (Well kinda, I've not seen the 4th movie yet...)

Quote from: Leon on 21 March 2023, 02:22:50 PMPretty much and we know how that ended!  (Well kinda, I've not seen the 4th movie yet...)

Don't bother, it is very disappointing.  A film too far you might say.

I don't know whether to be impressed or concerned about this kind of technology. 

Well both, but more the latter. AI art and what it can do is tearing up the art community, though most people I know are horrified by it. I have to confess I am too. Certainly the field of concept art is under extreme threat.

The speed at which this progressing is what's truly scary. By the time people cotton on to this it will be too late. It's not that AI is going to wipe out humanity Skynet style, it's that it's just going to take your job. If your work is mostly based on cognition, you're under threat.

The big thing lately is that in education its soon (if it isn't already) going to be impossible to tell if student essays are student written or AI written. A fool would think "great now AI can do my work for me", a smarter person will realise "oh they'll do away with essays as assessments". A forward-thinking person might worry, "If AI can write the essay, can it do the job I am being educated for?" It may not be immediately obvious that it is a threat today, but you have to think about where it will next year, the year after and so on.

Now is the redundancy of most people a disaster? That's not clear. Maybe we'll be the overseers of AI, picking out the parts of its work we find useful. We might be embarking on a world of plenty where work is no longer needed (and even that might be good or bad). Or we might be going down a path of wealth flowing ever upwards from the majority to the few at the top who control the technology.
What irritates me its that the whole AI bandwangon barrels on without much discussion of the ethics of it or even awareness.

Quote from: clibinarium on 21 March 2023, 10:01:37 PMWell both, but more the latter. AI art and what it can do is tearing up the art community, though most people I know are horrified by it. I have to confess I am too. Certainly the field of concept art is under extreme threat.

The artwork thing is interesting too, some of our guys have played about with it here.  I think the easy solution for that would be to monetise the technology so that the artists are getting a share of it, a bit like Spotify or iTunes but for art. 

Artists could submit their images into the database and allow use of them in AI art, but then receive a payment everytime their artwork is used.  That would mean charging for the AI technology but that seems to be the best way to approach it, rather than pirating everything on the web in one mass trawl.