
Interested in history, languages, software, dogs, dog software, and weird stuff from the internet. Swagless normie. Email me: [email protected]
Madison, Wisconsin
Joined December 2022
- Tweets2.2K
- Following865
- Followers280
- Likes5.2K
Pinned Tweet
Replying to @Scholars_Stage @zenahitz
everyone wants to philosophize and no one wants to live in a barrel
When I'm talking about "tokens" on here I am actually referring to Culver's Kids Meal Ice Cream Tokens.
25
My proposal for the new Bond series is that you set it in the near future where public key crypto is done for and Bond basically cannot use a phone/computer. Semi-futuristic/semi-1960's.
The Riemann hypothesis has fallen. What's the contingency plan if factorization is next (thus undermining public-key cryptography)? Where is this being discussed?
Also: people made encryption schemes that people couldn't break; who is using AI to invent better cryptography?
35
Many things (Why Anthropic continues developing something they think is extremely dangerous, Why Anthropic is valued at 1.5 trillion, etc.) are based on the premise that AI is winner-take-all.
At the risk of embarrassing myself, I don't think that's true.
Interesting remarks Anthropic made in a meeting with the sell side:
- Anthropic has the stability and the better talent that open source lacks.
- The AI market is winner take all. As long as we execute properly, we will take all of it.
- The contract with SpaceX will not be extended beyond the middle of next year.
- Developing Opus 5.5 cost more than $10 billion.
Readers added context they thought people might want to know
The remarks are unverified second-hand notes from a Barclays analyst call on Anthropic meetings, not official statements. Poster corrected $10B cost to full Claude 5.5 family; filings show SpaceX deals extend to 2029 but terminable after early 2027 on 90 days notice.
x.com/jukan05/status…
fourweekmba.com/ai-anthropics-…
sec.gov/Archives/edgar…
My biggest hot take right now is I couldn’t be more bearish on the Anthropic IPO for the long term. Like obviously they’ve built an insanely good model but as a business I just don’t see the value in AI accruing to the frontier labs. To be clear not investment advice and I have no position (Nor do I have positions in any other stock for that matter). Succinctly though my issues with it are the following:
-Distillation is never going to stop and candidly its pretty obviously to me legal to retain the data from a product that you pay for. There is no serious long term way to stop this that I can see and at best its a constant game of whackamole. As long as Claude is a model you can use via either app or API it will keep happening and distillation is a fraction of the cost of the training.
-Even if you buy into they’re going to stop offering models and eat SaaS companies and just offer an entire catalog of products this issue doesn’t really go away. If it costs them 0 to produce software then it does for everyone else too.
-They’re not positioned at all for robots and even if they massively expand headcount to go after white collar services businesses that just then balloons costs and they look like every other company plus STILL there is the competition issue that doesn’t go away. Further they don’t fully understand these industries and its entirely possible they end up whiffing and losing tons of money doing some huge rollout.
-If you think that future of robots is really where the value in AI lies I would tend to agree but they can’t product their own robots so they have to install basically their OS into other people’s devices. The parallels with Apple and the ios store and apps is too strong here and the actual hardware manufacturers always end up winning.
-AI is an insanely powerful piece of tech but the current paradigm of brute force scaling compute tops out pretty quickly and its not clear it can ever get actually efficient especially with all this competition.
-The current LLM based paradigm is strong and can go further but its not clear to me that more “intelligence” is actually the issue or the bottleneck in a lot of industries. It would take a whole new approach imho to make any kind of serious leap and I don’t think machines will ever get to a true human level because (and this sounds wishy washy and hand wavvy hippie style but bear with me) without discernment that comes from context and emotions you’re capped at usefulness. Pure logic only gets you so far. Yes I'm a Yann fanboy in this area. No I don't care.
-Their entire financials are predicated around this notion that they are profitable if you don’t count the cost of training the models. This is the equivalent of saying like Toyota would be profitable if they didn’t develop a new car every year. The idea that a lab can simply stop producing new models is farcical to me because the switching costs between companies is near 0.
-Further as the AI bubble pops or deflates compute is going to get cheaper over the long term plus as manufacturing and data centers continue to ramp and get more efficient the cost for compute keeps dropping in the medium and long term. On the surface this would seem to help them but it really just opens it up for more competition that can train more specialized models easier and faster too.
-Even versus the other AI companies their nanny state safety culture is going to bite them hard. I was early to hating how hard Anthropic polices what you do with Claude but more and more people are noticing this and pivoting. Cutting off entire industries like cybersecurity except for the privileged few is going to bite them. This isn’t even to mention how they silently degrade models to try and save money which pisses off the power users that drive like 80%+ of usage.
All this to say yes I think there is a business here and obviously they’re making a ton of revenue for now but its priced to be this dominant take over the world company and I simply don’t buy it. Maybe I'm totally wrong though.
1
20
I agree with this to some extent but I also think people do want to work fewer hours, for fewer years, with lower stakes, doing more interesting things.
The connective tissue between AI-boosters who think AI will take all the jobs—necessitating UBI—and neoliberals who want to deregulate and then redistribute handouts down to the have-nots, is a drastic under-estimation of the extent to which the human species is a working animal.
31
one problem with looking to science-fiction for positive visions of the future is that narrative forms generally elide anything mundane. if progress comes mostly by making the mundane easier and better, it won't show up in TV because TV characters already don't do the dishes, etc
1
45
A graph I'd like to see: The largest amount of capital any one person has given over to AI to manage on its own, over time.
And how much does it explain this graph?
teapawt.net/alth0u/status/21228629…
2
29
The cutesy name is a perfect encapsulation of the unseriousness of these people.
Ahead of Anthropic’s IPO, a rare coalition across AI safety, open source, e/acc, science, and law has signed the MisAnthropic Statement. Read the statement or sign at: misanthropicstatement.com/
1
40
Amy's Organic Foods still not AGI pilled enough, printing "best by 12/2029" on their vegetarian chili
25
just to be clear: there are strong incentives to post if you are an AI maximalist, an AI doomer, an AI hater, etc. other perspectives are likely under-reported.
"we will create a world where there is a single stock, ours, but otherwise things are basically the same. do you want a share?"
Investing in Anthropic presents a strange paradox. If the ants are wrong about their commercial prospects, there's no reason to buy their shares. But if they're right about the transformative or potentially catastrophic implications of their technology, the world could change so radically that owning shares becomes meaningless. So the investment thesis requires threading a very narrow needle: believing Anthropic will be enormously successful economically, but that its leadership is fundamentally wrong about the consequences of that success. And even in that middle scenario, it's unclear how much upside remains for investors at today's valuation. Also, they are all vocal members of a crazy human extinction apocalpyse movement.
2
47

















