Urban75 Home About Offline BrixtonBuzz Contact

14 year old boy kills himself following obsessive use of AI chatbot

There was a pretty good LRB piece on AI chatbots recently:

This bit:

Some of the pessimism surrounding AI chatbots stems from a belief that humans, like computers, can be hacked: that our normal programming can be bypassed with the right combination of words. To invest language with such power is charmingly old-fashioned, like Socrates decrying the corrupting power of poets. It’s also an incomplete model of human behaviour, which fails to account for the myriad cues we use to judge the veracity and intentions of social actors.

I'm not so sure about. Even before the advent of conversational AI, social engineering techniques have been good enough for scammers to not just make a tidy living by selecting for and then exploiting especially vulnerable victims, but to create an entire industry based on it. That seems close enough to "people hacking" in my opinion.
 
AI is going to be terrible for photographers.

I saw a service where you can send some really bad photos (eg pics caught mid-sneeze, one eye closed etc.) and it churns out a set of professional-looking shots ideal for work profiles etc. for v little money.
I suppose the opposite point of view might be that it provides cheap photographs to creative types who can use them themselves.
 
Sadly not one with articulated arms and a spinning thingy in a recessed chest cavity.
Any of those where you press buttons and get a coffee. Like the one that’s been beeping incessantly for at least three weeks in my local Tesco Express.
Ah yes; I see. Not sure I’d call them a robot myself but don’t know much about the whole AI thing.

I thought chat GPT was an online doctor app for video appointments for quite a while (not even a joke)
 
The other day I wished to purchase a cup of tea from a stall in a hospital, and I was unable to do so, because the machine that make the tea needed to be re-booted. I remember kettles.
I remember hearing about a certain building where they had doors with a fancy electronic security system that people have to swipe in and out of, which had the unfortunate side-effect of meaning that if the network had issues then not only did the printers go down but also it became impossible to open the fire doors.
 
Really? Not because “intellectual property” is an obvious fiction that suited capitalism at a particular point in time?
I mean it has less protection than something like automatic driven cars.

Even though progress in the tech is comparable your going to see 6 fingered people in your artwork even long after most most auto cars are better than the average person.

That would be true even if we had better artist rights.
 
Really? Not because “intellectual property” is an obvious fiction that suited capitalism at a particular point in time?
It's no more a fiction than any other form of property. The notion that any thing or concept might belong to a person or persons is only in our heads ultimately, respected out of convention and rules that are also all in our heads. I'm not sure there's an important difference between 'intellectual' property and property involving physical things.

What is property? It is a claim to an exclusive right to use something. What that something is can be anything at all.
 
I mean it has less protection than something like automatic driven cars.

Even though progress in the tech is comparable your going to see 6 fingered people in your artwork even long after most most auto cars are better than the average person.

That would be true even if we had better artist rights.

I don’t think any of this is the case. I think the fingers thing will be a long-forgotten glitch well before self-driving cars are any good.

Except in Norfolk, obv.
 
As with everything else, I imagine it's usage will benefit those best placed to take advantage. There might be some surprise packages within that framework, but the upwardly curving inequality will be boosted and celebrated.
 
I suppose the opposite point of view might be that it provides cheap photographs to creative types who can use them themselves.
Well I guess. Except it's all stolen. And they're building extra energy infrastructure cos it takes so much to do.


But yeah, it can write a substandard CV and do weird pictures where everybody's hands are fucked.

Oh and it's running out of trainable data cos it's already nicked everything decent. Which may be an issue.

And companies like Open AI are running $5 billion a year losses.

But cheap photos maybe?
 
As with everything else, I imagine it's usage will benefit those best placed to take advantage. There might be some surprise packages within that framework, but the upwardly curving inequality will be boosted and celebrated.
tbh the way it is being used by some people as a substitute for people is scary on many levels, not least that it is so much more diminished than a person.

A lot of these things are up in the air still, but I suspect that it will not become significantly less diminished. AI does not reason. It does not think. It does not understand. That's built-in. It can give an incredibly convincing show of being able to do all of those things, right up to the moment when it fails.

I won't speculate on the details of this particular awful case, but I will give a prediction that the moment it fails is likely to cause emotional devastation to anyone who has become attached to it. And the better it becomes, the harder it is to fail and the harder that failure will be felt.
 
My favourite story out of all of this is Scale AI. They are run by the "world's youngest billionaire". They classify training data for AI - such as self driving cars or the military. They're valued at an obscene amount of money.

Guess how they do it?

wiki said:
In 2017, Scale established Remotasks as its in-house outsourcing agency. It set up over a dozen facilities in Southeast Asia and Africa to train thousands of data labelers. Remotasks has been established as a separate brand for client confidentiality according to Scale. Early employees say this was done to make Scale's strategy less obvious to competitors and shield the company from scrutiny.[2]

Remotasks was successful in controlling costs and by mid-2019, Scale's margins had recovered to 69%.[2]

According to a 2022 study by University of Oxford researchers, it concluded Remotasks met the "minimum standards of fair work" in just 2 of 10 criteria.
 
Back
Top Bottom