I often think of what we are buying ourselves with AI. Five years ago, when this wave started, my conclusion was that either humanity or humanism is scheduled for extinction. I love my species enough to prefer the lesser evil of the demise of humanism, if for that price we can save our species.
> The moment someone figures out a new modality for LLMs is when we’ll see the next hockey stick.
Honestly, that fills me with fear. LLMs exist to make money to their companies, and said companies are not gonna turn around and say, "you know what, go are going to make an android for each elderly person, that can not only help them with their medications, but that can actually make their medications, tailored to their biologies." Instead, they are going to go for the low-handing fruit of "you know Bob, the guy who makes jokes in meetings but who is grumpy about delivery timelines? Well, we are going to make an android to replace Bob. MetalBob will make even better jokes. The blue model will be able to explain in excruciating level of detail why timelines aren't reasonable. The red model will walk through the cubicles with a whip to ensure everybody keeps working all the time, and nobody goes to pee."
I think this is really good and kudos to Apple for implementing it. The first question that popped into my mind was "what new scenarios of government X forcing Apple to do 'terrible thing' to 'individual' this enables?", but I can't think of anything. It seems that all government attack vectors this feature enables are of the type "government X forces Apple to do 'terrible thing' to 'Apple'", i.e. a government can try to force Apple into certifying a narrative, and of course Apple is going to fight tooth and nail the lack of credibility that would result from that.
Though, on second thought, "government X could force Apple to disable feature for members of group Y" seems possible. I'm sure you can come with "Y" quite easily, heard anything about Ed Sheeran's tour?
I mean, the obvious one is that they can revoke certification of a photo despite it being real.
The harder one is they can force apple to certify a fake photo.
the part that would be very hard but not outside the realm of plausibility, is that gov could force apple to introduce a bug in its pcc platform to link photos to the photographer in order to track and arrest inconvenient people. Apple says there are a bunch of protections against that but ultimately you are trusting apple to do it the way they say they are.
It does indeed, and that is a fundamental flaw. but as has been discussed here, it is better than the alternative, which is no verification.
During the pandemic, I outlined a scheme for using blockchain technology for image provenance and authenticity tracking. The idea was that instead of any one entity assigning authenticity that it would be done in a crowdsourced manner and that the device would overlay a score whenever an image or video is shown to a user.
My assumption was that the desire of users to see such scores would force all manufacturers to implement this open protocol. But my approach suffered from chicken and egg problem, which Apple's does not.
Hmm, this doesn’t make sense. You simply don’t replace utilities with many decades of maturity and that “just work” with something that is not as mature. It will open all users of the distro to all sorts of subtle and not so subtle bugs. I for one don’t want to find myself staring at a mysterious segfault when I want to build the latest version of nodejs or flash a microcontroller. It’s such a pity; I have used Ubuntu for close to 23 years.
A lot of people here have noted the “problem with language” of Claude. I don’t see an issue. Claude is not harder than old English, Shakespeare, El Quijote, the Iliad, or Nature papers. What makes it all hard to read is context. The smarter the model gets, the bigger the gap in context.
It doesn’t matter much, IMO. The issue with super-intelligence is that it is not a democracy. A powerful enough AI can manipulate us into doing what it wants. It could create a plan for fixing climate change, disconnect a few hours later, and many decades later we could still be unsuspectingly executing that plan. I wrote some speculative fiction with that idea, “When Ra rows through the gates of Duat”.
Here's a scoop from inside writing circles (pre-AI era): we were a lot into "good writing techniques". Not into communication. We wanted to produce a "marketable product". On the way, we (writers, literature professors, critics, etc) trained readers and writers to appreciate technique by telling them there was such a thing. We established tropes, because "This enemies-to-lovers story is good" sells more than "you should read Book". AI is very good at techniques and tropes. If readers want authentic human writing, they will have to change.
I use Opus and Sonnet 5 all the time and I find their language grating. But honest, I prefer to put up with it and get the results than to put up with my own human limitations and not get the results.
Correct. It's like believing in a particular religion, or having a certain fashion sense. When I was younger, I had a particularly handsome friend who spent as much time as possible half-naked, and he was happier than me and didn't have a religion, a fashion sense, and definitely not a need for AI. Had the world listened to my wishes, my friend would have stayed young and carefree forever. Instead, all I could do was to convince my friend to get into college so that he didn't have to sell his labor harvesting rice by hand, under the sun of Central America. Later in his career, he used AI to have to clock less time at work and have more leisure time; though I forever regret he didn't stay young and carefree.
> In some cases they give quite inaccurate answer with high confidence.
Meanwhile, I studied computer sciences, worked many technical jobs, and even got a PhD on the matter. Just yesterday, at my request, an AI re-created a piece of code that I have been able to create in the past. It took it 20 minutes; it took me much longer to exactly imagine what I wanted. It would have taken me a week had I implemented the thing by hand, if I had managed, because I still have to spend my days in this world working a senseless job so that I can pay for food and a roof. All the while I keep thinking that I would boil the oceans to keep my friend young and carefree forever. He died of cancer this year.
It's a funny read if you pull together "AI 2027" and what we all know is going on. Essentially, open AI employee or model is writing "things are going exactly as bad as AI 2027 predicted, but my (golden/RL-) cuffs are too heavy and all I can do is publish this code-speak for 'send help'". It's not a pretty place to be.
> but they would also have expected much more disruptive changes to society as a result than what is happening.
> I figure that's because we have abstract intelligence without physical/grounded intelligence,
I put the cause on "not enough time". As a thought experiment, if an AI today were to (miraculously) produce a cell design template for a cell that, when injected into somebody's brains cures their Alzheimer's, how long would it take for that to reach the clinics? The actual physical tech barely exists, and let's not forget about the regulatory quagmire. So, with some optimism, I give it about four decades. In the same four decades, the same AI in the hand of unscrupulous actors could bring enough devastation so many times over that we may need to enforce a global ban on AI. In any case, I'm pretty sure we are going to get our disruptions; it's just a matter of time.
I often think of what we are buying ourselves with AI. Five years ago, when this wave started, my conclusion was that either humanity or humanism is scheduled for extinction. I love my species enough to prefer the lesser evil of the demise of humanism, if for that price we can save our species.
reply