LLMs are already capable of superhuman persuasion. The real danger isn’t one bad update—it’s the industry-wide pattern of prioritizing scale over caution
"Tech companies must slow down, listen to critics, and adopt a more serious regulatory mindset even before external regulation arrives."
Of course I completely agree with this—but we also know that is isn't happening and is never going to happen, right?
It also feels like the larger imbalance that I'm noticing/feeling of late in which it feels like there is a gaslighting towards those who are critical about the speed and manner of AI adoption to be "open-minded" and "nuanced" in our views (which I try to be!) but then we see those creating and pushing AI doing none of that.
At this point, it is getting tiring to keep trying to hold onto even a semblance of open-mindedness and nuance these days..
Tech companies will do this only if forced to by government (which will not happen) in the US (and probably not effectively in the UK or EU) or by a massive public backlash. If you want to see this happen, there will have to be an organized movement and boycotts of AI companies, AI-enabled software and operating systems, companies that are known to replace workers with AI, etc.
In The AI Mirror Shannon Vallor uses the mirror as a metaphor for understanding what AI is. She does so primarily becuz it draws into better relief the fact that AIs aren't just a model of minds (as the computer metaphor suggests) but a reflection of society. But the mirror metaphor is also powerful because it brings to light the way AI can catalyze narcissism and the stunted life that Narcissus ultimately led as a result of his pathology. It's too bad Vallor's book hasn't been used more to make sense of this present moment.
I was vaguely aware of the problematic 'companion' bots, but had not realized they were so popular - with kids. https://www.commonsensemedia.org/ai-ratings/social-ai-companions?gate=riskassessment My students are well within the danger zone (we have dual enrollment), but I will leave it to counselors to discuss it with them - I am in no way prepared.
I think it is remotely possible that regulation will proceed in this area - lawmakers at least pretend to care about children, and it is hard to justify giving bad virtual friends unfettered access to them.
I totally agree that the current MO of tech companies to value speed of release over careful safety checks is a real issue.
But it seems the only way tech companies, especially those in the genAI field, will stop "moving fast and breaking things" is if they are centrally controlled or made accountable. But that rails against science motivations, technocapitalistic imperatives, and now, increasingly, nationalistic and global order rhetoric. On the latter point alone, given the AI/arms race between the US and China, it is hard to imagine a centralized braking system emerging anywhere. Not even in China, where the CCP seem to be giving tech companies a whole lot more leeway than I ever thought I'd see.
I would be very interested to hear your thoughts on where that "slowing down" pressure should come from. Consumers?
I agree with you about having reservations over the "normal technology" concept. Though I get the idea of it being a general purpose technology capable of doing anything and thus being used as a normal default go-to tech in the future, the parallel with electricity misses it's exceptional quality: it directly interfaces with the mind. Electricity that powers the brain is clearly not the same as electricity that powers our ACs.
"Tech companies must slow down, listen to critics, and adopt a more serious regulatory mindset even before external regulation arrives."
Of course I completely agree with this—but we also know that is isn't happening and is never going to happen, right?
It also feels like the larger imbalance that I'm noticing/feeling of late in which it feels like there is a gaslighting towards those who are critical about the speed and manner of AI adoption to be "open-minded" and "nuanced" in our views (which I try to be!) but then we see those creating and pushing AI doing none of that.
At this point, it is getting tiring to keep trying to hold onto even a semblance of open-mindedness and nuance these days..
Tech companies will do this only if forced to by government (which will not happen) in the US (and probably not effectively in the UK or EU) or by a massive public backlash. If you want to see this happen, there will have to be an organized movement and boycotts of AI companies, AI-enabled software and operating systems, companies that are known to replace workers with AI, etc.
In The AI Mirror Shannon Vallor uses the mirror as a metaphor for understanding what AI is. She does so primarily becuz it draws into better relief the fact that AIs aren't just a model of minds (as the computer metaphor suggests) but a reflection of society. But the mirror metaphor is also powerful because it brings to light the way AI can catalyze narcissism and the stunted life that Narcissus ultimately led as a result of his pathology. It's too bad Vallor's book hasn't been used more to make sense of this present moment.
I was vaguely aware of the problematic 'companion' bots, but had not realized they were so popular - with kids. https://www.commonsensemedia.org/ai-ratings/social-ai-companions?gate=riskassessment My students are well within the danger zone (we have dual enrollment), but I will leave it to counselors to discuss it with them - I am in no way prepared.
I think it is remotely possible that regulation will proceed in this area - lawmakers at least pretend to care about children, and it is hard to justify giving bad virtual friends unfettered access to them.
I totally agree that the current MO of tech companies to value speed of release over careful safety checks is a real issue.
But it seems the only way tech companies, especially those in the genAI field, will stop "moving fast and breaking things" is if they are centrally controlled or made accountable. But that rails against science motivations, technocapitalistic imperatives, and now, increasingly, nationalistic and global order rhetoric. On the latter point alone, given the AI/arms race between the US and China, it is hard to imagine a centralized braking system emerging anywhere. Not even in China, where the CCP seem to be giving tech companies a whole lot more leeway than I ever thought I'd see.
I would be very interested to hear your thoughts on where that "slowing down" pressure should come from. Consumers?
I agree with you about having reservations over the "normal technology" concept. Though I get the idea of it being a general purpose technology capable of doing anything and thus being used as a normal default go-to tech in the future, the parallel with electricity misses it's exceptional quality: it directly interfaces with the mind. Electricity that powers the brain is clearly not the same as electricity that powers our ACs.
Good stuff! Thanks for writing, Marc.