Sam Altman's Fall from Grace | Hacker News Reader

Sam Altman's Fall from Grace

garymarcus.substack.com

30 pointspasswordoops15 comments

I'm not really sure this is Altman's "Fall from Grace" moment, per se, because I'm not sure it's truly about him, even if it seems to be entirely his fault. I do think this has a chance to really sour everyone - including lawmakers - to AI companies and their wanton ways. The "just train on anything you can find" policy will surely come into question now. To quote a tweet quoted in the article:

> I think [Scarlet J] humanized the entire story of these companies stealing from everybody.

It was absolutely a risky move, and even if they're "innocent" of using her voice, the PR damage is done. It was a bold gambit for attention, and now we have to see the outcome. I think it's too bad too, because we've seen a ton of cool uses of LLMs and this "new age of AI" playing out. Is it a bit of hype? Yes. Is there useful tech there too? Absolutely - and I don't want OpenAI to ruin it by foolishly trying to give their CEO a cute cinematic moment.

Not directly related to the central thesis, but it's annoying how Marcus promotes the narrative of OpenAI thriving on theft of IP ... and then, every time he wants to quote someone, he posts a screenshot instead of linking the source.

give him a break, he's at least 3 pivots in on what criticism of OAI he actually wants to stick with - it must be hard to keep them all straight

Writing separate articles on all the reasons to be weary of OAI is not "pivoting". It's listing all the reasons.

To wit: weren't we 6 months from AGI over a year ago? Where's Q*? Where's ChatGPT5 for that matter?

> 6 months from AGI over a year ago

where did they say this?

> Where's Q*

These are internal techniques. As you might have noticed, nobody is publishing.

I remember the articles Gary Marcus wrote early on, which is why I know how incorrect he is. His pessimism about capabilities was severely misplaced which is why he has pivoted to copyright. Now he's even reposting people saying that Sam Altman doesn't care enough about extinction risks? His critiques aren't even self-consistent or coherent.

Again, not pivots. In fact here's a recent sampling of criticisms:

May 15, Safety issues: https://garymarcus.substack.com/p/chaos-and-tension-at-opena...

May 6, Tensions with MSFT: https://garymarcus.substack.com/p/microsoft-and-openais-incr...

May 5: sama as grifter: https://garymarcus.substack.com/p/the-sam-altman-playbook

yep, all of those articles are from May 2024. But I remember what he was saying in 2022 as I'm sure many other people on HN do - saying that the x-risk is ridiculous, these systems are fundamentally not very capable, etc. etc.

Now he's posting approvingly of people saying Sam Altman is imposing extinction risks on us all.

I definitely agree that Sam Altman is much less popular now than in the past.

Think this is inevitable given the current politics of tech. In my view, it seems obvious that tech has done a lot of good in the world, but this seems to rapidly becoming the minoritarian position.

The politics around tech nowadays seems extremely characterized by ressentiment. Traditional targets of popular sentiment like east coast tycoons & financiers have been supplanted by tech workers and CEOs as can be seen by the rise of the 'tech bro' narrative (despite the fact that most tech workers I know are by far the least bro-y of the professional classes.... they are more on the autism spectrum than doing typical bro-y things associated with finance culture).

I take solace in the fact that this mostly seems to be a phenomenon among the chattering class. When I leave DC, LA, and NYC, people mostly just seem really impressed with the progress being made.

He tweeted "HER". That has nothing to do with politics. It was so easy NOT to do that.

I think a useful enhancement of the motto "move fast and break things" would be "move fast and break things, but consider the PR consequences".

which absolutely don't matter. The dumbest thing you can do is worry about what Twitter/HN or the decel crowd thinks. (Google fell in that trap and still recovering from it)

People hate success. People hate success even more if it threatens their livelihood. openAI has launched a direct assault on 'experts', especially experts who aren't hands-on. So, their natural response is to attack the figure head of the AI revolution.

Twitter / HN although have many smart people, if you are addressing a mass market, you can completely ignore their opinions.

So, go Microsoft -- integrate AI directly into the OS.

go Sam -- built the biggest, baddest LLM model out there

Google, if you still haven't learnt the lesson of ignoring haters and taking risk, you have only yourself to blame.

Don't let 100 loud voices come in your way to serve 8 Billion customers

Sama: "Can I use your likeness in my new product?"

ScarJo: "No"

Sama: "LOL just kidding I'm going to do it anyway, suck it "

Every normal person: "sama overstepped."

Fanboys: "STOP BEING SUCCESS HATING ANTI-TECH LUDDITES!"

the fundamental thing is that Sky doesn't actually sound that much like her voice....

e: and after writing this I just looked it up, apparently it doesn't use her voice at all? not shocking to me, as again they do not sound that similar to me

his primary mistake was writing 'her' which gave the lawyers an in

This is classic HN/Twitter take who are absolutely clueless about the law/copyright.

Producer : I'm making a movie X. I wish Harrison Ford would play lead.

Harrison Ford : Suck on it.

Producer : Well this dude comes close to Ford. Let me cast him and make a movie anyway.

The fact that supposedly smart people are overly cautious and are heavily influenced by celebrities / hollywood take on this subject is proof why people like Sam Altman, Elon Musk succeed. Their competition is clueless / risk-averse

Reply on news.ycombinator.com