Let's talk about Artificial Intelligence

[rgmwa]

The major AI developers in the US recognise the risks and don't seem to be averse to some kind of regulation or constraint on development, but none of them wants to be the first to slow down because they're all in competition with each other and have huge sums of borrowed money tied up in development with very little return on investment to show for it so far. They can't afford to stop and seem to be hoping the government will step in and do something, which the current administration certainly won't. Another reason they won't slow down is that unless there's an international agreement, which will never happen, they will be worried that China or Europe will get the upper hand.  I think we're on a rocky road with great potential for positive benefits but also lots of opportunity for things to go off the rails. I saw a report today that the Houtis were using Claude to design them a new missile until Anthropic got wind of it and stopped them. 

 

[nomadpete]

20 hours ago, Jerry_Atrick said:

there has to be a global organisation and agreement on AI proliferation

There already is. There is a small number of very wealthy tech moguls (independant of political pressure, of course) looking after all that.

[Jerry_Atrick]

1 hour ago, rgmwa said:

The major AI developers in the US recognise the risks and don't seem to be averse to some kind of regulation or constraint on development, but none of them wants to be the first to slow down because they're all in competition with each other and have huge sums of borrowed money tied up in development with very little return on investment to show for it so far.

Yes.. but by my paralleing AI to guns, regulation is required to force it to slow down. However, emerging news is: https://www.independent.co.uk/tech/sam-altman-elon-musk-ai-safety-concerns-b3049217.html

 

Which basically says there is a need to slow down. The other problem is you have China and no doubt Russia and other non-aligned countries who will not slow down any time soon.. so, again in an analogy of weapons, you're only as strong as the effectiveness of what you've got. There has to be some sort of verifiable global non-proliferation agreement. 

 

Ultimately, like everything else in a free market, it takes government regulation to manage the risks. 

Edited 16 hours ago by Jerry_Atrick

[rgmwa]

1 hour ago, nomadpete said:

There already is. There is a small number of very wealthy tech moguls (independant of political pressure, of course) looking after all that.

True, but I was hoping for a more responsible group to manage it.

 

[Marty_d]

Yes - these models are being built by billionaire tech bros, as a cohort possibly the most amoral set of scoundrels as exists in the USA.

They are completely detached from reality. They have no concept of how normal people live.

This AI race is simply a game to them. Why do they care if their company "wins" this race? They already have more money than they could spend in a hundred lifetimes.

[Jerry_Atrick]

54 minutes ago, Marty_d said:

They already have more money than they could spend in a hundred lifetimes.

I have to admit, I am not normally against people accumulating vast amounts of wealth. Again, that is based on an economic structure where the wealth did trickle down at least to those who participated - the labour force. However, the rules are changing. While I am sure if you asked one of the LLM/generative AI solutions to build you your own AI competitor, they would reneg, there is nothing in the core model that would not allow them to do it. In other words, in terms of product development, it is almost a perpetual motion machine already.. Of course, ti still requires people, but disproportionately fewer for the output - and significantly so. 

 

On that basis, trickle of wealth down will be slower than an old fella's pipe without the magic tablets they take. So, on that basis, again, the wealth will have to be forced to trickle down by outside regulation - so at some stage a) the clamps have to be put on or b) there is a point at which that wealth is taxed. Of course there will be cries that it is paper wealth and if they cashed it out it would impact the share price of investors which includes pension funds. Of course, it doesn't have to be cashed out. The government can automatically take a stake in the companies that exceed a valuation / employee ratio or something like that. 

[Marty_d]

 

https://www.abc.net.au/news/2026-09-14/irish-reunification-natural-trump-says/107148712

 

Despite the text, the link is actually about how Trump thinks everything is ok with AI and it's just "very negative" people who worry about it.

So now I'm really concerned.

[old man emu]

It is interesting that the people doing the AI development work are now issuing warnings that the AI horse is kicking at the corral gate, thretening to bolt from all control. You know there is something amiss when the boss of Anthropic is one of those calling out warnings. However, we know that it is corporate greed that is ignoring these warnings.