801–810 of 1001 posts
Frequently we'll hear from Amodei that AI is 6 months or 12 months out from replacing Software Engineers. Or we'll hear that their model is safe when regulators step in, but suddenly it's too dangerous and hacked its way towards its goal when other companies claim the same.
What they have in common with Ed Zitron is that they are both trying to sell you something.
> February 2025: "I will keep writing this stuff until I’m proven wrong." Wrong (Zitron continues to write despite repeatedly being proven wrong) This was a good one.
Hilarious, but he never actually said he'd stop writing, just that he'd continue writing until proven wrong. And provably, he did. He wrote until then, and continued after. Since he never stopped writing, he met the challenge. He's now free to stop writing whenever he wants and still not fail that statement. ;)
Everything everyone is predicting has almost zero value. Nobody owns a crystal ball. Right now there could be someone working in a garage somewhere that could shake the entire LLM world to the core with a unique insight and innovation. Think about the world before the 2017 "Attention Is All You Need" paper. Did anyone predict that paper, what preceded and followed it? Nope. Same case now. Maybe the word "prediction"…
Even if you believe they can get cost to launch 1KG down to 100 bucks... (current falcon heavy is $1550) you still have the problem with all those solar panels.
1GW of space AI requires... 1 GW of solar to power it.
Current manufacturers of space solar is 1-2 MW &.. space solar aint cheap. $100 per watt. They estimate each spacex will be 250 kW. So.. $25M just for the solar for each satellite. Yeah you can use less efficient / cheaper solar, but then your weight goes up.
Also... Ground AI datacenters haven't nailed down how quickly GPUs depreciate. Accountants claim 6 years while CRWV says it's longer and most realists say it's shorter. It's based on how quickly GPUs will improve. Which is QUICKLY.
In space you'll have to depreciate GPU, all supporting hardware including the satellite, electronics, solar panels, and launch costs. On the ground it's just the GPU & you can even sell the old one down.
ground based AI needs power. $FRVO is an interesting ticker if you believe in their technology. Just announced a deal with google yesterday.
Earlier quoted context omitted.
> You can do that, but then you’re no longer discussing his predictions. You’re discussing your predictions, and your own positioning. Depends if you care about the "prediction" part or if you care about the assessment of the situation (regardless of date). If someone in 2000 said "the subprime mortgages market is a bubble and will blow no later than 2003", they got the prediction wrong, but their assessment would be…
Predicting things 5 years too early is often as useless as not predicting anything. You can say "AI will be able to _____" and be right 99.9 times out of 100, but the question is when. You can say "The AI market will go to 0" and be at least directionally right eventually. But none of it matters if you get the timing wrong.
It's not the predicting that's important to me, it's the assessing.
I don't care if it will do X in 2 or 5 years.
I do care if something IS "X", and we should correct course to not get the bad outcome associated with X (whenever it might come) or not.
Earlier quoted context omitted.
He made a blog post, literally last month, and it said that AIs have no use outside of coding. It just shows he's done zero research on the things he talks about all day. Radiologists are using them, ad firms, artists, translators, law firms, auditors... It's hard to think of a white collar firm not using them.
One thing I've noticed is that since LLMs came out, scientific papers with poor English have essentially disappeared. Just one of the many ways that AI is changing the world. People are using AI in all kinds of fields.
Things in general I think he's right about: 1. Revenue if anthropic and openai is unlikely to grow to the high levels they need to pay for their commitments. Many of their heavy users (coding) will eventually offset a lot of usage to more efficient and cheaper open weight models. I know of people in a company I was at that spend thousands of dollars a month on tokens. I am sure that what they're using it for can be s…
I saw a commit in the repo at work that edited a few config files and a dictionary.txt file. "co-authored by Fable"
Things in general I think he's right about: 1. Revenue if anthropic and openai is unlikely to grow to the high levels they need to pay for their commitments. Many of their heavy users (coding) will eventually offset a lot of usage to more efficient and cheaper open weight models. I know of people in a company I was at that spend thousands of dollars a month on tokens. I am sure that what they're using it for can be s…
> I am sure that what they're using it for can be substituted in large part by way cheaper models. I saw a commit in the repo at work that edited a few config files and a dictionary.txt file. "co-authored by Fable"
Things in general I think he's right about: 1. Revenue if anthropic and openai is unlikely to grow to the high levels they need to pay for their commitments. Many of their heavy users (coding) will eventually offset a lot of usage to more efficient and cheaper open weight models. I know of people in a company I was at that spend thousands of dollars a month on tokens. I am sure that what they're using it for can be s…
Take this one example of a user with 15 Codex subscriptions ($3k) generating $60k in API-equivalent usage per month: https://hraness.com/writing/my-girlfriend-asked-me-why-i-hav...
“there’s a once in a lifetime discount happening at the OpenAI Intelligence Depot, and I brought 15 shopping carts.”
We have to understand current usage with this behavior in mind, there are tens of thousands of people just like this author who are intentionally generating as much usage as possible on their subsidized plans because they feel compelled by some need to get free intelligence.
The only reason these expensive models are generating so much usage is because they are so heavily subsidized. Pragmatic users will shift to cheaper models which will hurt per token revenue, yes, but huge volumes of usage is going to just disappear because there isn’t the demand when it isn’t being subsidized. Less revenue per token and less tokens.
https://tokscale.ai/leaderboard a small sample of just 2k users have generated over $100m of API-equivalent usage while paying closer to just $1m.
Earlier quoted context omitted.
Hilarious, but he never actually said he'd stop writing, just that he'd continue writing until proven wrong. And provably, he did. He wrote until then, and continued after. Since he never stopped writing, he met the challenge. He's now free to stop writing whenever he wants and still not fail that statement. ;)
Are you a software tester?
Earlier quoted context omitted.
In my experience, people who make fun of AI's ability to eliminate white-collar work are exactly the same people who are white-knuckling it, hoping they can make another $300k next year to buy that Rivian. People who don't need a salary look at the situation more objectively and, in general, can see that a lot of white-collar work is in peril.
I think that is tied to the general trend that people see themselves as far less replaceable than others. And I’m not sure that people who don’t need a salary are exactly objective about it - unless you mean people who live off the state or something, those are people who need their capital reserves to continue to appreciate in order to sustain their lifestyles, and are thus relying on the destruction of white collar…
What if (almost) everyone is right, that their work really is more difficult to automate than others realize?
This could be the lesser minds problem. Based on it I am doubtful most knowledge workers are as immediately replaceable as everyone suspects.