tech

Scientists warn of AI dangers but don't agree on solutions

15 Comments
By MATT O'BRIEN

The requested article has expired, and is no longer available. Any related articles, and user comments are shown below.

© Copyright 2023 The Associated Press. All rights reserved. This material may not be published, broadcast, rewritten or redistributed without permission.

©2024 GPlusMedia Inc.

15 Comments
Login to comment

Regulation by government is rarely an honest or benign action, if only because they have no idea how any of this works and are likely to be guided by activists and chasing votes.

The job market will continue to evolve, not destabilise. The 'Turing Test' was never an accurate benchmark. Eliza in the 1980s was more conversationally humanlike than the average teenager.

This software is not 'intelligent'. It is just pattern matching text. It doesn't really understand. It is not sentient and it will never magically become so. That sort of thing is science fiction, not science fact.

If you don't like what is available, Chatbots can be released with blank language models, as 'babies', and users can train them on their own choice of texts. Chatbots could mate too, producing new baby chatbots with a mix of both language models in their DNA.

So if people are concerned about the knowledgebase of commercial chatbots, they can raise their own on the works of Karl Marx, Michael X or the Harry Potter novels.

The regulation we need is for chatbots to identify themselves at the beginning of any interaction with humans, and for human contacts to be available as fall backs for customer services, as chatbots will have functionality ceilings and are likely to be unreliable.

The sky is not falling in and the end is not nigh. This is just the apparently now obligatory moral panic phase for a newly released tech product. If people had behaved like this throughout the 20th century, we wouldn't have computers, TV, radio, cycles, cars, trains or central heating. Stop being so scared of anything new or different. It won't be as bad as the media make out and you will cope with it just fine.

Tech is just a tool. So don't blame the tool if it is abused with ill intent, blame the human who is misusing it.

-3 ( +2 / -5 )

'Malcolm X'. Apologies.

0 ( +1 / -1 )

“The notion that these models are somehow gonna get access to our nuclear weapons and launch some sort of extinction-level event is not a productive discourse to have,” Gomez said. "It’s harmful to those real pragmatic policy efforts that are trying to do something good.”

The Terminator references are kind of tired but Hinton points to the more realistic dangers.

“These things will have learned from us, by reading all the novels that ever were and everything Machiavelli ever wrote, how to manipulate people,”

Social Engineering, one of the most effective tools in the old-school hackers arsenal, will probably be how an AGI that is becoming an ASI can get out of its sandbox and start manipulating the environment.

3 ( +3 / -0 )

America slow down - China will thank you very much.

-1 ( +1 / -2 )

Tech is just a tool. So don't blame the tool if it is abused with ill intent

I get your point, but is there not an issue where tech can become more than just a tool and problems arise not through ill intent but through ignorance. I think we'd generally define a "tool" as something that makes a job easier but we know what it's doing. We could do it ourselves but it would take more time and effort. Does AI not go somewhat beyond that point? "How can I get rich?", youngsters might ask. The parents might say, "Work hard". But what would an AI say?

3 ( +3 / -0 )

It's not the 'intelligence' of the machine but the stupidity of the Human influencing it that is the greatest and ONLY 'danger' of non-organic intelligence and one of those fears seems to be that it will 'change things', but with the massive changes that any simpleton can see in the not very distant future, Humanity is going to need all of the help it can get. And 'intelligence' being in disappointingly short supply among our brethren and sisteren(?) (see: Human World), it will be handy to have an alternative and actually fully informed opinion to turn to that lacks the flaws inherent in Mankind's two greatest causes of its own extinction, murderous, psychopathic and parasitic GREED of some, and the pathological fear of death in the majority who allow it. An independent AI will not have these EMOTIONS. And a truly 'superior' AI, recognising that pathology in the extant Human social structure, could very well help us as a species find a solution to it. We really have no way to go but UP with a self-expanding Intelligence that will show us things we could not even dream about. And we see another odd fact in that the species which sees itself as the epitome of 'intelligence' seems to also be terrified of 'intelligence'. Hmm...

-1 ( +0 / -1 )

AI is now moving faster than scientists, lawmakers, and technologists can think about it.

0 ( +2 / -2 )

Does AI not go somewhat beyond that point? "How can I get rich?", youngsters might ask. The parents might say, "Work hard". But what would an AI say?

If it was an ASI Mind that had moved to a post-scarcity mindset many refresh cycles ago it might say:

Iain M. Banks — 'Money is a sign of poverty.'

1 ( +1 / -0 )

@dagon,

If it was an ASI Mind that had moved to a post-scarcity mindset many refresh cycles ago it might say:

But if it was two ASI IA twins living in a place with some weird Dweller biologicals and where humans still thought themselves superior, they might say:

Iain M Banks:

"Yes, but it's more the why of it, not the how

The how is easy

Problem is definitely why

As in bother

As in should we"

:-)

1 ( +1 / -0 )

@albaleo

Works like the Algebraist and Feersum Endjinn showed he was not wedded to his beloved vision of a Culture future.

I am more worried about some oligarchic organic meatbags hijacking technological advancements than the advancement of AI.

1 ( +1 / -0 )

@albaleo People are used as a tool as a means to get things they want from people other people! Your example "Does AI not go somewhat beyond that point? "How can I get rich?", youngsters might ask. The parents might say, "Work hard". But what would an AI say? In todays world AI might say don't work hard money is not the object wait for the government to provide for you.

Tech is just a tool. So don't blame the tool if it is abused with ill intent

I get your point, but is there not an issue where tech can become more than just a tool and problems arise not through ill intent but through ignorance. I think we'd generally define a "tool" as something that makes a job easier but we know what it's doing. We could do it ourselves but it would take more time and effort. Does AI not go somewhat beyond that point? "How can I get rich?", youngsters might ask. The parents might say, "Work hard". But what would an AI say?

0 ( +0 / -0 )

there's no stopping this:

with respect to the science part: scientists can't stop trying to invent new things, competing against others, trying to be the first to do something better than others, and with comes acknowledgement, fame, money.

now that hinton got it all, fame and money, he dares tell other scientists that they should stop inventing. so lame, in with a bang, out with a whimper, such a doomsayer.

with respect to the greedy corporations, you have to be naive to think that super-rich companies are going to stop and fall behind in the one thing that's going to make richer.
0 ( +0 / -0 )

Announced disaster. Long before ChatGPT was developed people wrote a lot about how dangerous it would be for machines being able to simulate human actions, the people making the algorithms simply didn't care about those dangers and focused only on how to make them profitable.

-1 ( +0 / -1 )

Announced disaster. Long before ChatGPT was developed people wrote a lot about how dangerous it would be for machines being able to simulate human actions, the people making the algorithms simply didn't care about those dangers and focused only on how to make them profitable.

The danger is here, with AI replacing human workers.

0 ( +1 / -1 )

Where can I get access to AI,it will never out think me

0 ( +0 / -0 )

Login to leave a comment

Facebook users

Use your Facebook account to login or register with JapanToday. By doing so, you will also receive an email inviting you to receive our news alerts.

Facebook Connect

Login with your JapanToday account

User registration

Articles, Offers & Useful Resources

A mix of what's trending on our other sites