> peak power consumption is about 200W for the Mac Studio with M3 Ultra, and 300W for the Mac Studio with M5 Ultra — a fraction of the power used by a common toaster.
Your argument is persuasive and well illustrated. I think the problem is the intro paragraphs come off as too certain of catastrophe which, when juxtaposed with your claim that "smarter people than me have been working on this", makes it sound like you don't actually believe they're smarter than you. The rest of your essay feels fair and not judgmental.
I do believe they're smarter than me, but sometimes very smart groups talk themselves into ultimately impractical solutions because they're all smart. Sometimes you need a dumb guy to come in and say "are you sure this is right?"
I believe you are sincere. But "is about to be a huge, costly, global train wreck" lacks the nuance of "are you sure this is right?" and will rub some people the wrong way.
Edit: I'm not suggesting you should have written it any differently. I think you made the right choice to be a bit provocative because it grabs the attention that's needed.
I do believe this. But that doesn't mean I can't be convinced otherwise by a good argument. The point of this post is to see if anyone has a great counterargument to change my mind.
I think you're basically right about this: if you're going to ask an ecosystem to bear the costs of a breaking change, you should be asking yourself what will be bought for the price you will pay.
For users it's a set of scales that hangs in balance. On one side is the disruptiveness of the change to them, and on the other is the promise of what they will gain once on the other side.
If there's a problem with the current thinking it's that it proposes to trigger a big expensive breakage that we'll be paying for for 10 years, but there hasn't been any serious discussion of what else is broken about git that needs to be fixed to git to continue to be relevant for the next 10 years.
Most discussions are pre-constrained by the idea that ecosystem wide breakages are impossible.
In theory it's good for me if the git ecosystem splits compatibility in half with no obvious benefit. That just makes it easier for me to waltz in with a third backwards-incompatible migration path, one which actually offers eye-popping kinds of benefits in return for the cost of the breakage. That's why I'm competing with you in the race to replace Github. It's just that I'm also racing to replace git, so I assume you'll be in some amount of trouble if I should succeed.
I can't tell if Khan is being realistic when she says laws against “unfair or deceptive [acts]” or “unfair methods of competition” apply to AI labs. Her 2-sentence justifications aren't enough to convince me. But maybe it's sufficient for state AG's who want to start investigations.
She is a former FTC Chair so she knows what she's talking about, although her approach to enforcement was more expansive than others. The FTC would investigate, likely along the lines she summarizes, build a case in writing, and refer it to the Justice Dept, which would actually prosecute the case. Not going to happen under this administration, but Khan is saying it's entirely possible that such an enforcement action could be brought to bear, now or later. Personally I hope she gets her job back at some point and does it herself.
I would point out that someone having been FTC Chair does not inherently mean that they know what they are talking about, as demonstrated by the current FTC Chair.
I certainly wish for this to be a productive line of attack against the AI labs, but I'd like to read a more thorough argument showing that it would hold up in court. (It's academic anyway; we'll all be paperclip'd by the time there's a change in administration and a prosecution can be brought and worked through appeals.)
that's certainly fair - it doesn't necessarily follow. her specifically, being an expert before and having spent time as chair though, does tell me she is on target here.
She isn't above demagoguery and bullshit slinging so I wouldn't treat her as infallible. Any more than the idiot with an axe to grind against Backdoor in a failed attempt to bolster a career.
> I can't tell if Khan is being realistic when she says laws against “unfair or deceptive [acts]” or “unfair methods of competition” apply to AI labs. Her 2-sentence justifications aren't enough to convince me.
As I read it, her whole point is that those laws apply to everyone, including AI Labs.
She isn't saying AI Labs are in violation of those laws (though she is implying they should be scrutinized for possibly being so), she's just saying that if they do violate those laws... well, those laws are already on the books and AI Labs aren't special unicorns for which existing laws somehow don't apply.
Cohere's disagreement seems mostly about who makes the rules. From the essay (which does seem significantly written by AI, sadly):
> The rules around AI are getting written either way. What's still open is whether they get written by a group anyone can join and with evidence anyone can check, or by a handful of companies in a room with the door shut.
Of course none of the AI companies will suggest that they should be accountable for the illegal actions taken by agents they host, but I feel like that could be a very effective "pacing mechanism", and addresses the concerns brought up by this essay about fairness and transparency.
> Of course none of the AI companies will suggest that they should be accountable for the illegal actions taken by agents they host, but I feel like that could be a very effective "pacing mechanism"...
The problem is that the potential liabilities arguably vastly exceed what these companies can absorb. It would be interesting to see if any major insurer wanted to step up and insure them.
On the plus side, in the worst case scenarios, we won't have to worry about damages because nobody will be left to pursue them.
Just look at Meta's recent settlement. The minds and mental health of countless young people harmed. And they'll pay $17 billion, which is about one month's worth of revenue.
The legal system today does not serve the purpose you're wanting it to.
Then the law is simple, no insurance? No product. No need to slow down development. Anyone who can make a good product should develop as fast as possible.
English common law has been one of the greatest legal inventions of all time for business and accountability. Under common law a company that releases a product that has unintended consequences is liable for those.
So naturally companies want to pretend this is somehow different.
There's the punctuation and word choice tells. And the overall voice. But let's set that aside because some human writing shows similar traits, or perhaps a human used an LLM to "fix the grammar". So I'll point out another tell that this article is pure slop:
This article is overflowing with links, numbers, and graphs. When a human writes that densely, it takes effort to incorporate all those details. So the human only does that if each detail contributes to their argument. In this article it's mostly irrelevant. It's LLM filler to pad out an article that really has nothing to say beyond its (also LLM-scented) title.
The blog's About page says Max "writes" about transportation, but I'm skeptical he's even read his own slop here, or he would have presumably fixed some of the worst nonsense sentences before posting it. I doubt he even had the courtesy to ask Claude to run a fact-check on its output. Curse anyone who wastes my reading time like this.
Thanks for the link. His decades of tech columns were so well-served by his skeptic curmudgeon style. There's danger in bringing that persona to all topics, I now see, shown by this regrettable chapter of his blogging life.
I'm glad this worked for your family, but in my kids' experience this is too heavy to use as balance bike. By the time they were comfortable hoisting their de-pedaled 12" / 5.4kg bike off the ground, they'd nearly outgrown it. They preferred a cheap hand-me-down balance bike which was half the weight (the upscale ones with brakes and pneumatic tires are self-defeatingly heavier).
Just as ChatGPT is now called "Chat" by the kids (go ahead and finish grinding your teeth, I'll wait...), if Antigravity takes off then it'll get an irksome nickname like Antigrav or just Grav.
(Emphasis added.)