AI imposters
Everything online may soon enough be fake, posted by artificially intelligent computer algorithms (AIs) that are programmed to engage (and annoy) people. What we encounter online may have been altered by AIs to the extent that normal humans aren’t quite sure whether they’re dealing with something real from real human beings. First, Samuel Hammond describes a vision of the future that sounds plausible enough, without being excessively terrifying:
[Having] a personal super assistant in every person’s pocket will do far more to empower [AI on location].
Indeed, within a decade, ordinary people will have more capabilities than a CIA agent does today. […] You’ll be able to replace your face and voice with those of someone else in real time, allowing anyone to socially engineer their way into anything. Bots will slide into your DMs and have long, engaging conversations with you until [they sense] the best moment to send [a] phishing link. […] Relationships will fall apart when the AI lets you know, via microexpressions, that he didn’t really mean it when he said he loved you. […] Public comments on new regulations will overflow with millions of cogent and entirely unique submissions that the regulator must, by law, individually read and respond to. Death-by-kamikaze drone will surpass mass shootings as the best way to enact a lurid revenge. The courts, meanwhile, will be flooded with lawsuits because who needs to pay attorney fees when your phone can file an airtight motion for you?
By his reckoning, we get there gradually rather than all at once, but with the effects spread out everywhere. His vision isn’t a simplistic one of evil AI suddenly taking control with a singular secret purpose.
Next, Arnold Kling asks how humans and human systems will deal with that amount of attempted fakery. He says the answer will be in setting up good AI that has the sole job of looking out for AI made with criminal purposes in mind.
So if I were working on AI alignment, I would try to treat it in part as a cybersecurity problem. I would try to develop systems to detect when hackers are trying to trick the AI into doing bad things. Then I would implement protocols that cripple those hackers. If the hackers are trying to use the AI for criminal activity, the protocols would involve working with the police.
What I am left wondering is how all the digital fakes and forgeries end up affecting us as a species on the whole, from the level of the individual on up, with our constant weakness for fooling ourselves, for falling for our own biases and prejudices. A subservient AI will probably just try to feed us more of what we already want to believe. And if we don’t like what the AI in our pocket tells us, we can delete it and replace it with something that tells us what we’d prefer to hear—an optimized digital yes-man.
Talk about fake news and alternative facts! The ever-present human weakness for self-deception, the constant desire to join a tribe so as to partake actively in things larger than ourselves: Will AI help hold this in check, or accelerate the tendencies such that the perpetual fights of the past few years look like a children’s dress rehearsal before real, knock-down-drag-out live-action role-play?
Ugh. So much for avoiding catastrophism today… Then again, maybe that is the ultimate human tendency: To foresee a future of doom and cataclysm no matter what. Don’t believe me? Even if you aren’t prone to believing we’ll incinerate ourselves and our world intentionally or accidentally in the next decades or so, give or take: our best, state-of-the-art science knowledge says our sun Sol will eventually balloon out to swallow up this planet, and then the Universe itself will fly apart until all the embers wink out of existence in eternal cold emptiness.
Those of us of a more spiritual mindset will take solace in the thought that we’ll be together in some sort of afterlife we can’t really imagine. And why shouldn’t we believe so, really? Fears of total annihilation have been with our species forever—and none have yet come true. We just aren’t that good at predicting things—especially not about the future.
+++
I will be out the next few days, just able to check from time to time. I’ve got daily postings in the pipeline, though. So don’t touch that dial!
Hello all...been a bananas day at work , first minute I had to get here
Last night my TV wouldn't work, today it apparently mysteriously starting working again...lol...
I already have a laptop that is going to go blue screen of death on me eventually , don't need another techy failure...lol
I don't know, I don't talk to many people online that I already don't know to some extent...And I think I have good instincts about authenticity most of the time, I can almost always detect trolls or fakery...I also don't believe we are that self destructive...OTOH, by that point I could be dead or not doing much online at all anyway...lol...
I do like some of the exciting part of this exploration/develoment...it can be used for good also
I am off on Friday, Thursday is our two person office party, though I am trying to get him to actually leave the building and go to the BBQ place down the street....there is alcohol here though also...lol
I am disappointed my immediate family had to reschedule Christmas Eve again, third year in a row...my brother has to work 11 to 11 that day and we don't want to do it without him, we are going to try for New Year's Eve, early afternoon...of course, that means for Christmas Eve I have nothing much to do ( other than stuffing for the turkey Christmas Day maybe) until later at night when we go to Rick's sister's house
Excellent, if potentially terrifying, post MG. I appreciate all the work you consistently put into this little Dispatch home away from home!