It's Openai's World, We're Just Living in It

(We were joking about it just last week because my partner asked my eldest what was the Power Point he was working on and he said, "Whats Power Point?")

Maybe some are too young to remember the great migrations from/to MySpace, MSN, ICQ, AIM, Skype, local alternatives like StudiVZ, ..., where people used to keep in contact with friends. Facebook was just the latest and largest platform where people kept in touch and expressed themselves in some way. People adding each other on Facebook before others to keep in touch hasn't been a thing for 5 years. It's Instagram, Discord, and WhatsApp nowadays depending on your social circle (two of which Meta wisely bought because they saw the writing on the wall).

If I open Facebook nowadays, then out of ~130 people I used to keep in touch with through that platform, pretty much nobody is still doing anything on there. The only sign of life will be some people showing as online because they've the facebook app installed to use direct messaging.

No, people easily migrate between these platforms. All it takes is put your new handles (discord ID/phone number/etc) as a sticky so people know where to find you. And especially children will always use whichever platform their parents don't.

Small caveat: This is a German perspective. I don't doubt there's some countries where Facebook is still doing well.

When you realize this, you realize that a lot of other supposedly valuable tech companies operate in the exact same way. Worrying that our parents' retirement depends heavily on their valuations!

As Peter Thiel says: “competition is for losers”

The bigger problem is that there was never a way to move data between oracle->postgres in pure data form (i.e. point pgsql at your oracle folder and it "just works"). Migration is always a pain, and thus there is a substantial degree of stickiness, due to the cost of moving databases both in terms of risk and effort.

In contrast, vendors [1] are literally offering third party LLMS (such as claude) in addition to their own and offering one-click switching. This means users can try and if they desire switch with little friction.

[1] https://blog.jetbrains.com/ai/2025/09/introducing-claude-age...

It turns out that they can use the same prompt system for all of them, with no changes and still solve 5/6 IMO problems. I think this is possibly iffy, since people might have updated the models etc., but it's pretty obvious that this kind of thing is how OpenAI are doing their multi-stage thinking thing for maths internally.

Consequently if prompt systems are this transferable for these hard problems, why wouldn't both they and individual prompts, be highly transferable in general?

Changing the LLM backend in some IDE is as complicated as selecting an option in a dropbox for those who integrate such a feature. They are other scenarios where it might be a bit more complicated to transition of course, but that's it.

True, but that's not really applicable here since LLMs themselves are not stable, and are certainly not stable within a vendors own product line. Like imagine if every time Oracle shipped a new version it was significantly behaviorally inconsistent with the previous one. Upgrading within a vendor and switching vendors ends up being the same task. So you quickly solidify on either

1) never upgrading, although with these being cloud services that's not necessarily feasible, and since LLMs are far from a local maxima in quality that'd quickly leave your stack obsolete

or

2) being forced to be robust, which makes it easy to migrate to other vendors

That applies even when you switch models within a vendor, though.

The vendors have all standardised on OpenAIs API surface - you can use OpenAIs SDK with a number of providers - so switching is very easy. There are also quite a few services that offer this as a service.

The real test is does a different LLM work - hence the need to evals to check.

OpenAI turned that research into a product before Google, which is a huge failure on Google's part, but that's orthogonal to the invention of what powers modern models.

Also Claude Opus 4.1 runs multidimensional circles around GPT-5 in my view. The only better use case for GPT-5 is when you need it to scrape the web for data

Like Apple vs every other computer maker.

Their platform mostly just works. Their api playground and docs are like 100x better than whatever garbage Anthropic has.

I think their UX is way better, and I can have long AF conversations without lagging. I can even change models in the same conversation. Basic shit Anthropic can’t figure out (they can fleece their 20x max subscribers tho)

I think if they get the AI human interface right, they will have their iPhone moment and 10x.

Makes one wonder if Google will eventually sweep this field.

In the EU/UK you might not have rights to the memories right now, but you've rights to the inputs that created those memories in the first place.

Wouldn't be too hard to export your chat history into a different AI automatically.

No moat.

The competitors have not come even close to Google's level of quality.

With LLMs, it's different. Gemini/Claude are as good, for the most part. And users don't care that much either - most use the standard free ChatGPT, which likely is worse than many competitors' paid models.

Google Search would be moatless if not for the AdMob purchase.

Meanwhile bing search is actually a platform, and is what then powers other "search engines" (duckduckgo, kagi, etc...)

Edit: mixed up my dates claiming DALL E came out before GPT 3

(You can say default in various browsers and a phone OS and that's probably the main component but it's not clear changing that default would let Bing win or etc.)

can the world and tech survive fruitfully without AI? yes. can the world and tech survive without electricity and transistors - not really. the modern world would come crashing down if transistors and electricity disappeared overnight. if AI disappeared over night the world might just be a better place.

https://youtu.be/OYlQyPo-L4g

From 7th minute, truth about ChatGPT - 2% paid users.

In my experience he has a horrible response to criticism. He's right on the AI stuff, but he responds to both legitimate and illegitimate feedback without much thoughtfulness, usually non-sequitur redirect or ad hominem.

In his defense though, I expect 97% of feedback he gets is Sam Altman glazers, and he must be tired.