Hacker News Reader

Comment by omeze | Hacker News Reader

I seriously wonder why, given the Microsoft deal, OpenAI didn't just cede all product development and production-izing of their models to MS while they focus on research.

I know smart people work there, some friends included, but OpenAI was never a product-led startup, they were a research org. Ignoring the closed vs open debate for a moment: for all the talk of focus, focus, focus from YC, it seems unfocused to ship their models from a home-rolled platform when they could just get MS to foot the engineering and infra work. Is any of their platform stuff a real asset to OpenAI proper? Have MS do the cloud stuff, just pass them product design.

just my 2c from the outside.

View on HN

Because when you accidentally discover a golden goose, you don’t give it to Microsoft, you learn to cook eggs.

I had the same thought as you at first, and many people speculated they’d shutter ChatGPT in favor of the API being served to others. But while anyone can sell an API, making a consumer product people want is just too rare to give it up. It’s better to learn how to service consumers and satisfy the demand.

Everyone was quick to point out that the API costs of a chatGPT clone were a fraction of the cost of ChatGPT+. So why would they focus on research and API development? I worry that they were so successful with this LLM that they’ll forget to keep researching other stuff.

That's my worry, too. I'm reminded of the history of Intel, which at its founding decided that it would not have a research org (drop the R from R&D) and just focus on scaling FETs, because the founders learned that trying to do both meant that one would ultimately suffer.

Scaling FETs is not research?

It’s all research - it’s just that when you get a taste of the money, the trend is often to lean heavy towards more of it. Wasteful, frivolous curiosity often falls into the rear view.

There are not a lot of places where you can pursue "wasteful and frivolous curiosity". And even at those places, you would need to have earned the privilege.

What did OpenAI do wrong here? As far as I can tell they're guilty of requiring logins?

It's not requiring logins per se -- this is the second mishap I'm aware of related to their consumer product surface area (the first being users' chat data going to other users). Just seems like a distraction, if the goal is to do fundamental AI research.

I think you're missing the point: this isn't a mishap that has anything to do with OpenAI. If you are a company that has usernames and passwords, then your login information is for sale because your customers have malware on their computers. OpenAI didn't do anything wrong here besides have a ton of users, many of which had malware on their computers.

From the article:

> However, the ChatGPT parent company clarified the compromised login credentials were not the result of any OpenAI data breach. Instead, they were the by-product of commodity malware-based log harvesting.

> If you are a company that has usernames and passwords

> didn't do anything wrong here besides have a ton of users

Passwords aren't a trivial part of this. OpenAI offers login with Google/Apple/MS. Managing their own passwords as well wasn't a mandatory part of the product, but a trade off: they probably got more users, but also more complexity and risks.

Federation has a privacy risk, and I fully support non-federated logins.

But still no, a malware thst grabs passwords doesn’t mean you can blame a company for using passwords.

He didn't read the article. Headline and comment.

Microsoft is not doing so great in the security department right now.

>Microsoft comes under blistering criticism for “grossly irresponsible” security

https://news.ycombinator.com/item?id=36979532

you didn't read the article.

The stolen OpenAI credentials were stolen using Raccoon Infostealer (78,348), Vidar (12,984), and RedLine (6,773) malware variants.

Just my 2 cents.

Noob Thought on why OpenAI collaborated with MS:

1. I have found ChatGPT is very good at parroting back decent quality from learning from structured data. For example, learning reasonable code completions as well as answering technical questions. I find ChatGPT very good resource to have inside VSCode instead of switching to google. 6-7/10 I will accept GPTs output in Copilot.

2. My guess is that Microsoft's pretty chunk of profit is from selling Developer software(or so it is trying to do).

Put 1 + 2 together and you have a pretty compelling product play.

They have much larger plays and interests than developer software such as Office/Outlook. Also stops Google from making the same deal as OpenAI needed credits from someone.

OpenAI doesn’t make money from research.

They dont make money from authentication systems either

Yes, hence the MS deal (they had trouble raising money as a pure research org). Their research can be turned into a product people pay for -- they've done it! My question is: why did the engineers at OpenAI have to do that? Why couldn't they "toss it over the fence" to MS's Azure team and be happy with a revenue share, for example.

MS didn’t buy them, just invested shit ton of money and got a close partnership.

Don’t outsource your money making department to a big tech company. They’ll screw you over, even if they don’t mean to.

One reason for having their own consumer website is that having lots of data from actual users should be quite helpful for training. I believe they promised not to use the data from other companies that use their API?

It's also good marketing. For the people who use ChatGPT every day, they're a known quantity, unreliable as it is. Other companies need to show that they're better.

These are both similar advantages to what Google has with their search engines.

Chatgpt is programmer art. I don’t think they have actual designers or frontend devs working there.

I accidentally deleted my chat history because of the confusing UX.

https://news.ycombinator.com/item?id=36979532

Eh, MS isn't great engineering IMHO. It's good enough but nowadays tends to have flaky edge cases.

Reply on news.ycombinator.com