Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Operator research preview

openai.com

Post reply on HN
Print view View original

291–300 of 448 posts

Re: Operator research preview

#291

Post by reissbaker »

Volundr wrote: post #233
chipgap98 wrote: post #226

Earlier quoted context omitted.

Isn't there every reason to believe the cost will come down?

Is there actually reason to believe costs will come down significantly? I've been under the impression that companies like OpenAI and Google have been selling this stuff at well below cost to drive adoption with the idea that over time efficiency improvements would make it possible, but that those improvements don't seem to be materializing, but I'm not particularly informed in this so I'd love to hear a more informe…

The costs for OpenAI and Google aren't public, but if you look at the open-source models, inference is very cheap: for example, you can generally beat the public serverless prices by a factor of ~2 by using dedicated GPUs [1], and given that a 70b model costs about $1/million tokens serverless — and tend to perform similarly on benchmarks to 4o — OpenAI is most likely getting very fat profit margins at $2.50/million input tokens and $10/million output tokens.

The problem for them is making enough money for the training runs (where it seems like their strategy is to raise money on the hope they achieve some kind of runaway self-improving effect that grants them an effective monopoly on the leading models, combined with regulatory pushes to ban their competitors) — but it seems very unlikely to me that they're losing money serving the models.

1: https://fireworks.ai/blog/why-gpus-on-demand

Re: Operator research preview

#293

Post by freediver »

CaptainFever wrote: post #176
gordon_freeman wrote: post #155

What is fascinating about this announcement is if you look into future after considerable improvements in product and the model, we will be just chatting with ChatGPT to book dinner tables, flights, buy groceries and do all sort of mundane and hugely boring things we do on the web, just by talking to the agents. I'd definitely love that.

I would really love for Apple Knowledge Navigator to be real: https://www.youtube.com/watch?v=umJsITGzXd0 and I'm surprised that people don't bring this visualisation up more often.

This is something we'd like to build. It requires owning both hardware and software - you can not build this in the world of platforms with permissionless apps.

Re: Operator research preview

#294

Post by windowlessmonad »

sumedh wrote: post #280
windowlessmonad wrote: post #177

Earlier quoted context omitted.

Are our attention spans so shot that we consider booking a reservation at a restaurant or buying groceries "hugely boring"? And do we value convenience so much that we're willing to sacrifice a huge breadth of options for whatever sponsor du jour OpenAI wants to serve us just to save less than 10 minutes? And would this company spend billions of dollars for this infinitesimally small increase in convenience? No, of c…

> Are our attention spans so shot that we consider booking a reservation at a restaurant or buying groceries "hugely boring"? Dont be limited with these examples. How about Airline booking, try different airlines, go to the confirmation screen. then the user can check if everything is allright and if the user wants to finish the booking on the most cheapest one.

Airline booking is a solved problem. Google, Expedia, and many others have their hands on flight pricing and can show you comparisons of those in a single query. Takes 2 minutes. What is the value add of AI here? Making the experience feel like a conversation and at a hyper inflated cost and resource usage with the risk of hallucination? No thanks, solved problem.

Re: Operator research preview

#295

Post by mkagenius »

turingbook wrote: post #292

CogAgent: Open source alternative from China. Paper： https://arxiv.org/abs/2312.08914 Code： https://github.com/THUDM/CogAgent Model: https://huggingface.co/THUDM/cogagent-9b-20241220

Click3: open source alternative for Mobile operator

Code: https://GitHub.com/BandarLabs/clickclickclick

Re: Operator research preview

#296

Post by neom »

chipgap98 wrote: post #226
mplewis wrote: post #216

Earlier quoted context omitted.

And none of the competitors can make this technology profitable, either.

Isn't there every reason to believe the cost will come down?

The cost of what? Training a model or trained a served model? The cost of both benefit from economies of scale. If I had what openAI has I could imagine how to make it profitable tomorrow, because I could do that is why they HAVE to make it free without an account, to prevent anyone new from meaningfully entering the $0 to $20mth segment, they already know nobody can compete with the most advanced model.

If you look at their business strategy, it's top notch, anchor pricing on the 200, 20 sweet spot, probably costs them on average $5/mth to server the $20/mth customers, Take your $50m a year marketing budget and use it to buy servers, run a highly optimized "good enough" model that is basically just wikipedia in chatbot and you don't need to spend a dime on marketing if you don't want to, amazing top of funnel to the rest of your product line. I believe Sam when he says they're losing money on the $200/mth product, but it makes the $20/mth product look so good...

They're really playing business very well.

Re: Operator research preview

#297

Post by sumedh »

windowlessmonad wrote: post #294
sumedh wrote: post #280

Earlier quoted context omitted.

> Are our attention spans so shot that we consider booking a reservation at a restaurant or buying groceries "hugely boring"? Dont be limited with these examples. How about Airline booking, try different airlines, go to the confirmation screen. then the user can check if everything is allright and if the user wants to finish the booking on the most cheapest one.

Airline booking is a solved problem. Google, Expedia, and many others have their hands on flight pricing and can show you comparisons of those in a single query. Takes 2 minutes. What is the value add of AI here? Making the experience feel like a conversation and at a hyper inflated cost and resource usage with the risk of hallucination? No thanks, solved problem.

It's not a solved problem though, the final cost on the website can be different based on what options you select

Re: Operator research preview

#298

Post by 827a »

sumedh wrote: post #284
827a wrote: post #200

Every site that offers any service remotely interesting to humans will soon require a captcha to do anything.

Arent most of the SOTA AI models able to solve captchas?

One hopes that, even if they're capable of doing so, these companies will explicitly forbid it. Or, at worst, that these systems are traversing the web with a proper user-agent, and captchas/sites can infer its an AI from that.

Of course, when billions are on the line, and eventually when people are off running their own, ethics get murky. Longer term, I suspect we'll see a global push toward requiring cryptographically asserted human-ness through deep integration with hardware. Cloudflare (and, I'm sure, others) has done some significant work in this domain [1].

[1] https://developers.cloudflare.com/fundamentals/reference/cry...

Re: Operator research preview

#299

Post by rm_-rf_slash »

alach11 wrote: post #14

I don't know if I'm ready to hand over my grocery shopping (or date night planning) to an agent. But if pricing is reasonable, this could be a powerful alternative to normal RPA. Instead of hardcoding some automation using Selenium, this would be a great option for automating repetitive tasks with legacy business software, which often lacks modern APIs.

Grocery shopping is just a use case for people to wrap their heads around. Everyone has to eat.

If they demonstrated a big value add like automating CRM a smaller subset of professionals would be absolutely awed but most people would be scratching their heads wondering what it’s good for.

Post reply on HN