What 81,000 people want from AI

Investors want to see you use your own product, if they themselves don't feel the product is good enough to write their own announcement then investors would worry about their future.

And AI is still a product primarily aimed at investors and not consumers.

However, since we're frank here, I'd say I'll download the most recent release and be very careful about upgrading because I don't put much trust in projects co-created with LLMs. I know there is a full spectrum but I've seen enough and I don't have the resources to check where on the spectrum your project ends up. LLMs are a powerful drug and terribly hard to stop once you start.

True in the current state of LLMs, possibly not true forever if someone finds the magic bullet that turns the one-shotting (reliable) software dream that companies like Anthropic and Perplexity currently peddle into reality. Seems far-fetched ATM but the gains since GPT-2 have been very real.

We're quite a ways away from this though, even with Opus 4.6 and the like. And even further from it being part of Claude Code rather than some proprietary $1000/mo. closed-source solution.

As you say though, _if_ such a technology were to exist, it's Anthropic that holds all the cards, not random entrepreneur #25721 who is asking the Anthropic API the same thing that the actual customer could just be asking directly. At that point you're an undesirable middleman, not a business.

Then there's people who are "well intentioned", I guess, but lack the technical knowledge. A friend of a friend with no technical background is selling websites to companies that he writes with Claude. They look shiny, everyone's happy in the short run, but I don't doubt issues will come up down the line that someone will have to be responsible for. I'd personally feel like I was ripping people off doing this, but I think also Dunning-Kruger prevents you from knowing any better if you are the type of person doing this.

Then there's the whole B2B SaaS gang that are basically just producing vaporware and telling other people how to produce more vaporware. This is no different from crypto, NFTs etc. before it really. Just people trying to hustle others.

And then there's the whole clawdbot gang probably burning more in tokens everyday than normal people use in a month so they can sort 18 e-mails.

So yeah I mean you're right, there certainly is a subset of people who are using this ethically (as ethically as you can use LLMs but that's another story) to make some money on the side. Certainly not the majority though I'd say.

Even if true, no one is ever going to do that because of Dunning Kruger, so it's still not magically solving problems.

Okay how are we measuring this? We can't even quantify intelligence for humans accurately, let alone compare it to machines. Hell, we can't even really define intelligence.

I mean, humans can learn on the fly and progressively, and currently no LLMs are capable of that. Literally none of them, and no context doesn't count. So if that's the measure, then LLMs sit at a 0 along with rocks and twigs and humans closer to a 1.

Obviously that's not really the measurement, LLMs are quite good. But I don't think we can say, for sure, LLMs are a replacement for humans. They might replace some specific tasks, but humans are not a set of tasks. I'd still rather have 10 engineers than 0 engineers and 10 Claude Code licenses.

They're good at acting as a "reverse dictionary" like this where you give it a description of something, and it knows the word for it. They have approximate knowlege of many things.

This is a competition of public and private interests. A sick individual is going to lobby for tests until they discover the cause. From a public perspective, it might be cheaper to just let them die. AI is an advocate for the individual.

For the record, ChatGPT helped me diagnose a lifelong illness. I'm a new man now thanks to AI. Literally life changing. I had spent decades pleading for tests because no one could figure out the cause. I think a likely outcome here is not necessarily 10,000x more tests performed, but similar or even fewer tests, because the diagnosis success rate with AI is higher. It's not subject to bias. People tend to be more honest and reflective with their AI than they are with doctors. They get 5 minutes to give the entire case to the doctor. With an AI they can spend weeks debating and reflecting. This builds a case history far more detailed and accurate than anything we have in modern medicine today. Amplified by an order of magnitude because the AI can extract meaningful insights from the discussion.

In the very near future our AI will contact our GP for us. Soon after that, our GP will be our AI.

That was my take with the entire report which I think lends to an inherent bias within the data and stories. You have the entrepreneurial stories, then you have the ones where people are both impacted and receiving benefits.

The infographics and charts even call out how countries that are "first-world" with fewer safety nets are more likely to be in "survival" mode compared to countries with them.

The bit from George Carlin standup routine regarding how the poor are there just to scare the hell out of the middle class rings true in this reflection. Poorer countries accept their current realities and the feedback reflects the hustle. Richer countries with safety nets reflect the existential issues with previous industrial revolutions. Richer countries without safety nets reflect the fear that their efforts will be made "replaceable" by AI.

As for the rest - massive testing creating false positives - that is an issue of implementation and the errors introduced by humans, not data itself. If the process were in large part made more automated, it could screen for a larger panel of issues for less cost.

From my experience working deep in data and human factors - the issue in quantifying the root cause isn't reality, we live a shared experience in general. The issue is the data isn't good enough. What bugs us about it is the psychology that our perceptions are different enough to the degree that we will fight to prove an unknown.

The fact I'm young-ish and healthy looking, with good skin and hair, leads many doctors to outright dismiss me. Never mind my history of cancer and the undeniable fact that I am obviously not healthy. But I can also use the squishyness to my advantage. I talk confidently, I push back, and that works. It sort of short-circuits a lot of doctor's brains.

I have no clue what this would look like other than maybe an investment fund for people creating apps/businesses based on Claude tools.

You'll either need to freelance, or start a company (or maybe a co-op) to capture the new value created by your ability to leverage AI.

It won't be much different to when a company buys more CNC machines and the employees don't get any more money despite producing way more parts.

What I need instead is something that takes the burden off my entire society and gives them a breather. Universal health care to start. They could also use a higher minimum wage, and lower housing costs.

This is quite easy. Just optimize the models to do reviews and bug finding. This would make developers (who normally hate reviews) quite happy and let them do more coding, thus delivering more value and possibly earning more...

If they wanted to do this, they could put their models in a public trust for the public's access and benefit in research, education, etc. Then it could be licensed, pay a dividend like a sovereign wealth fund, etc.

Considering that they copy and train on the sum total of all human creativity, a public trust is something that would be in line with both the spirit, and first and fourth considerations, of fair use doctrine:

    1. the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes;
    2. the nature of the copyrighted work;
    3. the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and
    4. the effect of the use upon the potential market for or value of the copyrighted work.

So... not much benefit either.

AI chat bots will summarize the top N web search results as if they're fact, weaving them into seemingly coherent narratives, all while reassuring the user that their questions are really good and they're learning a lot.

It's like those recipe sites that have 5 pages of nice photos and background story and side tracks and whatnot as the author waxes verbose, so they need to put a 'Jump to recipe' button in so people don't just click 'Back' immediately.

Except this time for an article.

I can't tell if 'skip the junk' is good (junk can be skipped!) or bad (maybe this means there's too much junk on the page?)

What would you have done, in those circumstances though? Would you 'round up', overstating your case, like Anthropic did? Or would you ensure that you avoid the suggestion that you were misrepresenting the numbers?