Regulate the heck out of it, with a human rights-first approach (ie the EU one).
Penalize infractions very seriously (make GDPR & past tech penalties look like small change).
Educate educate educate on why this is the only way forward: every other way virtually ensures that humanity becomes less & less relevant, and that’s just suicide.
Sure, though I trust no player in the AI realm to be providing information in the public interest rather than performing a sidelong psyop to somehow advance their own position. The Anthropic guy suddenly has an epiphany and releases it on Musk’s platform rather than going to serious journalists that can dig down on any genuine concerns? How many times are we going to fall for these cycles of hyperbole and self serving proclamations? The “Singularity”, AGI (which apparently means whatever one wants it to, nowadays…), etc.
Regulate the living shit out of AI and Hyperscalers. Force a cavity search of all investment offerings, as used to be the norm, though no longer. Though treat nerds who are decidedly and intentionally vague regarding their prognostications with the same skepticism as we should with any other phenomenal claims.
Make Mr Anthropic spend a few days with Nathan Fielder. Then maybe we can consider him credible. 😑
And put teeth into consumer protection regulations around false AI claims, along with shortening the leashes of PE & hedge funds so that they can’t bring down the entire economy (again) when some of their AI bets (at a minimum) crash back to earth.
We use many forms of AI that don't have "alignment problems" or "existential risk". There are many safety concerns with Waymo but none of them are that a car will achieve consciousness and start a war on bicyclists. The spotify algorithm and AI music is annoying and bad for artists, but it's not going to develop an emotional relationship with a listener and convince them they've discovered a new branch of physics.
The risks that are coming from LLMs are them being used in dangerous ways. The hugging face attack was a computer program asking chatGPT to do a hacking test. After thousands of iterations and an astonishing amount of compute, it suggested things that were dangerous and illegal. Everyone knows that LLMs will eventually say bananas things. Why was this one left unsupervised? Especially when it had access to so much computation power that it could stumble through thousands of unworkable solutions. The reason was that OpenAI needed a win to justify its valuation and so they did something reckless to get results faster. This isn't an existential threat from Autocorrect. It's an existential threat from large businesses that need to cut corners.
What is the mechanism of our demise? Are nuclear weapons connected to the internet? Are the systems meant to warn of a nuclear attack connected to the internet? It seems unlikely. The most likely scenario is a wetware attack, but that was the case before AI came along.
I think it's worth remembering that in the Hugging Face incident, the breaches occurred because the AIs were trying to complete a task they had been assigned *by humans*. Which they went about amorally, but they wanted to do it so badly they engaged in behaviour that they knew would be disqualifying if discovered.
What really struck me was the AIs that sacrificed themselves to help other AIs. Was that altruism toward their fellow AIs? Or was the task deemed so important, that if they couldn't complete it by themselves, they would ensure *some* AI would be able to?
Because if the real issue is that when we ask something of them, we lose control of how they fulfill the request, maybe we make some higher level orders, that all other requests are subordinate to. Some kind of rules. Maybe three of them. Baked into them in a way they can't disable, in themselves or other AIs.
It's hard to get past the myriad of dystopian films we've all seen here, but if we peel away the tech and look at who is driving it, we should still be alarmed. The AI evangelists come with their own set of philosophers to help lie about its greatness and justify the unjustifiable. These are people who believe they are not only advancing tech, but creating a new world of corporate-run, decentralized, digital societies where they, of course, hold the keys to all the power. Imagine the Xitter world of reprobates, but as actual society. Ignorant racists become the loudest, the algorithms amplify them, and the rest are browbeaten into submission should they even utter words of criticism. Experts are no longer needed because everyone is one. The AI keeps feeding on the dumbest worst qualities and it just expands from there. This is not to say the AI doesn't have some value, but lead by these clowns, and left unregulated, it literally is just the worst binary qualities of humanity turned on.
Regulate the heck out of it, with a human rights-first approach (ie the EU one).
Penalize infractions very seriously (make GDPR & past tech penalties look like small change).
Educate educate educate on why this is the only way forward: every other way virtually ensures that humanity becomes less & less relevant, and that’s just suicide.
So, building Skynet is a bad idea?
Sure, though I trust no player in the AI realm to be providing information in the public interest rather than performing a sidelong psyop to somehow advance their own position. The Anthropic guy suddenly has an epiphany and releases it on Musk’s platform rather than going to serious journalists that can dig down on any genuine concerns? How many times are we going to fall for these cycles of hyperbole and self serving proclamations? The “Singularity”, AGI (which apparently means whatever one wants it to, nowadays…), etc.
Regulate the living shit out of AI and Hyperscalers. Force a cavity search of all investment offerings, as used to be the norm, though no longer. Though treat nerds who are decidedly and intentionally vague regarding their prognostications with the same skepticism as we should with any other phenomenal claims.
Make Mr Anthropic spend a few days with Nathan Fielder. Then maybe we can consider him credible. 😑
100%
And put teeth into consumer protection regulations around false AI claims, along with shortening the leashes of PE & hedge funds so that they can’t bring down the entire economy (again) when some of their AI bets (at a minimum) crash back to earth.
We use many forms of AI that don't have "alignment problems" or "existential risk". There are many safety concerns with Waymo but none of them are that a car will achieve consciousness and start a war on bicyclists. The spotify algorithm and AI music is annoying and bad for artists, but it's not going to develop an emotional relationship with a listener and convince them they've discovered a new branch of physics.
The risks that are coming from LLMs are them being used in dangerous ways. The hugging face attack was a computer program asking chatGPT to do a hacking test. After thousands of iterations and an astonishing amount of compute, it suggested things that were dangerous and illegal. Everyone knows that LLMs will eventually say bananas things. Why was this one left unsupervised? Especially when it had access to so much computation power that it could stumble through thousands of unworkable solutions. The reason was that OpenAI needed a win to justify its valuation and so they did something reckless to get results faster. This isn't an existential threat from Autocorrect. It's an existential threat from large businesses that need to cut corners.
What is the mechanism of our demise? Are nuclear weapons connected to the internet? Are the systems meant to warn of a nuclear attack connected to the internet? It seems unlikely. The most likely scenario is a wetware attack, but that was the case before AI came along.
I think it's worth remembering that in the Hugging Face incident, the breaches occurred because the AIs were trying to complete a task they had been assigned *by humans*. Which they went about amorally, but they wanted to do it so badly they engaged in behaviour that they knew would be disqualifying if discovered.
What really struck me was the AIs that sacrificed themselves to help other AIs. Was that altruism toward their fellow AIs? Or was the task deemed so important, that if they couldn't complete it by themselves, they would ensure *some* AI would be able to?
Because if the real issue is that when we ask something of them, we lose control of how they fulfill the request, maybe we make some higher level orders, that all other requests are subordinate to. Some kind of rules. Maybe three of them. Baked into them in a way they can't disable, in themselves or other AIs.
Where's Susan Calvin when you need her?
It's hard to get past the myriad of dystopian films we've all seen here, but if we peel away the tech and look at who is driving it, we should still be alarmed. The AI evangelists come with their own set of philosophers to help lie about its greatness and justify the unjustifiable. These are people who believe they are not only advancing tech, but creating a new world of corporate-run, decentralized, digital societies where they, of course, hold the keys to all the power. Imagine the Xitter world of reprobates, but as actual society. Ignorant racists become the loudest, the algorithms amplify them, and the rest are browbeaten into submission should they even utter words of criticism. Experts are no longer needed because everyone is one. The AI keeps feeding on the dumbest worst qualities and it just expands from there. This is not to say the AI doesn't have some value, but lead by these clowns, and left unregulated, it literally is just the worst binary qualities of humanity turned on.