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

111–120 of 448 posts

Re: Operator research preview

#111

Post by raincole »

jprete wrote: post #96
raincole wrote: post #78

Earlier quoted context omitted.

> but I always imagined some sort of standard, where apps and services can expose a set of pre-approved actions on the user's behalf I sincerely hope it's not the future we're heading to (but it might be inevitable, sadly). If it becomes a popular trend, developers will start making "AI-first" apps that you have to use AI to interact with to get the full functionality. See also: mobile first.

Why would developers do that? The developer's incentive is to control the experience for a mix of the users' ends and the developer's ends. Functionality being what users want and monetization being what developers want. Devs don't expose APIs for the same reason why hackers want them - it commodifies the service. An AI-first app only makes sense if the developer controls the AI and is developing the app to sell AI s…

Because Apple. Apple has the power over developers not the other way around, and it has shown quite strong interest in integrating AI into their products.

For example, by guiding your users to app instead of website, you immediately "lost" 30% of your potential revenue from them. On paper it sounds like something no one would every do. But in reality most developers do that.

Re: Operator research preview

#112

Post by xnx »

jsheard wrote: post #71
madeofpalk wrote: post #50

Earlier quoted context omitted.

> But probably more difficult to get app developers on board. That's it. The problem is getting Postmates to agree to give away control of their UI. Giving away their ability to upsell you and push whatever makes them more money. Its never going to happen. Netflix still isn't integrated with Apple TV properly because they don't want to give away that access. I'm not convinced this is the path forward for computers ei…

> I'm not convinced this is the path forward for computers either though. With this approach they'll have to contend with the agent running into all the anti-bot measures that sites have implemented to deal with abuse. CAPTCHAs, flagging or blocking datacenter IP addresses, etc. Maybe deals could be struck to allow agents to be whitelisted, but that assumes the agents won't also be used for abuse. If you could get Ch…

> With this approach they'll have to contend with the agent running into all the anti-bot measures that sites have implemented to deal with abuse

I expect many more sites to adopt login requirements. This has the added benefit of more tracking/marketing data.

Re: Operator research preview

#113

Post by Nevermark »

madeofpalk wrote: post #50
brap wrote: post #31

I don't know why, but the approach where "agents" accomplish things by using a mouse and keyboard and looking at pixels always seemed off to me. I understand that in theory it's more flexible, but I always imagined some sort of standard, where apps and services can expose a set of pre-approved actions on the user's behalf. And the user can add/revoke privileges from agents at any point. Kind of like OAuth scopes. Ima…

> But probably more difficult to get app developers on board. That's it. The problem is getting Postmates to agree to give away control of their UI. Giving away their ability to upsell you and push whatever makes them more money. Its never going to happen. Netflix still isn't integrated with Apple TV properly because they don't want to give away that access. I'm not convinced this is the path forward for computers ei…

This is classic disruption vulnerability creation in real time.

AI’s are (just) starting to devalue the moat benefits of human-only interfaces. New entrants that preemptively give up on human-only “security” or moats, have a clear new opening at the low end. Especially with development costs dropping. (Specifics of product or service being favorable.)

As for the problem of machine attacks on machine friendly API’s:

Sometime, the only defense against attacks by machines will be some kind of micropayment system. Payments too small to be relevant to anyone getting value, but don’t scale for anyone trying to externalize costs onto their target (what all attacks essentially are).

Re: Operator research preview

#114

Post by fassssst »

easterncalculus wrote: post #22

From the slide deck on the livestream: "[Operator safety risks and mitigations] Harmful tasks: User is misaligned" Looking forward to seeing some more of the examples for when openai considers their users as "misaligned", whatever that actually even means anymore.

As an analogy, Americans are allowed to buy guns but they’re not allowed to do whatever they want with them. An agent on the internet could be used for more harm than a gun.

Re: Operator research preview

#116

Post by 93po »

darioush wrote: post #35
easterncalculus wrote: post #22

From the slide deck on the livestream: "[Operator safety risks and mitigations] Harmful tasks: User is misaligned" Looking forward to seeing some more of the examples for when openai considers their users as "misaligned", whatever that actually even means anymore.

As the storyline unfolds "AI" seems to be code for "machine learning based censorship". Soon we will have home appliances and vehicles telling you about how aligned you are, and whether you need to improve your alignment score before you can open your fridge. It is only a matter of time before this will apply to your financial transactions as well.

drink verification can

Re: Operator research preview

#118

Post by sergiotapia »

moralestapia wrote: post #67
sergiotapia wrote: post #29

Earlier quoted context omitted.

https://github.com/bytedance/UI-TARS-desktop - I think it is proven there is no moat here. As much as there is a moat on "water" or "electricity" or "chicken breast". Intelligence will be sold for fractions of pennys.

Oh yeah, how could I forget about an obscure repo from a company that's getting banned from the US! It's simple, with trillions at play, if it's so easy to steal OpenAI's game, why has no one done it yet? Don't "argue" about it, just go and grab the money, it's easy, right?

Just dropped, 5x cheaper than Deepseek, which was already bonkers cheap. No moat.

https://x.com/deedydas/status/1882479771428544663

Re: Operator research preview

#120

Post by mynameisvlad »

celestialcheese wrote: post #20
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.

Locked behind their $200/mo plan - definitely too much for me with the accuracy they're showing.

For now, as a research preview. It isn't a stretch to think that it'll slowly be rolled out to their other plans.
Post reply on HN