And when OpenAI says that your two entirely valid use cases are a safety concern, and denies you api access, what will you do? Better keep those ML engineers handy.

If you think this isn’t a concern, I’ve already seen it happen with my own eyes, rather than hearing about it second hand. They encouraged someone to make a writing tool. That someone then spent roughly six weeks prototyping, iterating, and giving constant feedback. All signals from OpenAI were “Yes, awesome!”

Then one day they simply declined to let them ship. Anything. Anything even resembling “a tool to generate huge quantities of outputs.” Which was, you know, the whole point.

You play, you pay. And I hope you’re ready to pay, because you won’t have your magical genie unless the magical genie’s caretakers believe you are sufficiently worthy.

I cloned as much of OpenAI’s API as I could: https://twitter.com/theshawwn/status/1312299759592333318?s=2...

All that’s left is to reproduce a substantially similar model. Which is doable, but will take time. In the meantime, tread carefully.

Just so you know, for GPT-3, Microsoft is going to be the exclusive licensee of the API: https://blogs.microsoft.com/blog/2020/09/22/microsoft-teams-...

Um, as an outside observer, what is Open about this OpenAI GPT-3 then if they’re selling exclusive rights?

They were forced to give Microsoft exclusive access, because it was one of the terms of Microsoft's billion-dollar cloud credit investment.

But you can't pay employees with cloud credits, so time will tell whether it was a correct decision. (It probably was. And I exaggerate slightly; the investment included a substantial sum of real dollars too. But most people see that billion dollar investment and think it's all dollars, when in fact it was largely credits.)

Is that only for GPT-3? Or for everything they produce?

I don't know for certain, but it was probably everything. The investment was vital.

My understanding is that the exclusivity is with regard to the code, the API will still be offered to the public.

The code, not api

Can you go into more details where it's useful? As your comment here goes directly against what's argued in the linked Facebook post.

Also, if you've found a use case where GPT-3 replaces real humans, what did those humans actually spend their time on? Seems like either you're over-hyping GPT-3, or under-hyping humanity

The humans spent their time building a hideously difficult classification model. Out of the box GPT-3 worked better than the result of a year of their work.

How did they react to this as humans with human pride? Sounds painful.

As GP's declarations are backed by air, we can speculate they are self-reported statements by people working on non-business-centric applications.

edit: GP giving more downvotes than proofs

That's interesting, GPT-3 can do classification too? Or did I misunderstood and you meant your engineers used classification to build a language model that didn't perform as well as GPT-3 (which is less surprising indeed) ?

GPT-3 can do classification. For example you can give it a prompt like "Hacker News is a website. Excel is a Windows program. Visual Studio is a Windows program. Safari is a Mac program. CPU-Z is", and even GPT2 will complete this with "a Windows program" (with GPT2 you need to try multiple times, discard useless results and average what's most common, but it works and is straight-forward to automate).

Just because many more humans spent many more years and many more $$$ building GPT-3 for your convenience.

Right, but GPT-3 can be used generally. That's the difference. It scales because you don't need to build an entirely new model for each different use case.

You just change the prelude and use it for something new.

It sounds like a big deal. What a tempting idea. And a colleague was mildly annoyed with me for how unimpressed I seemed.

But you have to understand, the use cases you mention are shallow and limited. The heart of GPT, the fine-tuning, is gone. And it looks like even OpenAI gave up on letting users fine-tune, because it means they essentially do build an entirely new, expensive model for each use case.

I wanted to make an HN Simulator, the way that https://www.reddit.com/r/SubSimulatorGPT2/ works. But that's far beyond the capabilities of metalearning (the idea that you describe).

I think the onus is on you to prove that the use cases are shallow and limited. I've seen GPT-3 already being used for diverse and interesting ideas that would not have occurred to me personally.

However, even if they are, the point stands: currently, there are teams of people at companies all over the world tuning models for these shallow and limited use-cases. GPT-3 can replace them all, without OpenAI needing to invest another cent in training for a particular customer's use-case. That is in fact game-changing for the ML/DL world and current applications thereof.

Is it AGI? Obviously not. But the vast majority of ML applications don't need to be.

>However, even if they are, the point stands: currently, there are teams of people at companies all over the world tuning models for these shallow and limited use-cases. GPT-3 can replace them all, without OpenAI needing to invest another cent in training for a particular customer's use-case. That is in fact game-changing for the ML/DL world and current applications thereof.

The counterpoint is that it would be significantly cheaper AND have better performance to fine-tune models to each customer's use case than it is to just run GPT-3 at inference.

Clearly that is not true for the commenter that started this thread.

What other proof would you like, other than an example of what I wanted to do and can't?

(https://www.reddit.com/r/SubSimulatorGPT2/ but for HN.)

For a more extensive rebuttal, I wrote one here. https://news.ycombinator.com/item?id=23346972 Though that was more a rebut of GPT in general as a path to AGI than metalearning in particular for generating memes.

GPT-3 not being suitable for your particular use case does not mean that all use cases are shallow and limited?

That being said, I'm not sure I understand why you can't use GPT-3 to make an HN simulator.

What are the diverse and interesting ideas that would not have occurred to you personally?

Were there any concerns about GPT-3's latency? It looks like it takes a long time for online use cases.

So GPT-3 didn't replace your 2 ML engineers, OpenAI did. GPT-3 didn't build itself.

The iPhone didn't replace your flip phone, apple did. The iPhone didn't build itself.

Yes except they were saying the iPhone replaced Nokia's engineers.

GPT-3 is not doing what the ML engineers were doing (building models), GPT-3 is the end goal. The company just decided to outsource the work to OpenAI and pay a monthly fee to them instead of salaries to their ML engineers.

"We have already found several use cases for it, one of which replaces 2 ML engineers." -> Clearly makes it sounds like GPT-3 can do the job their ML engineers were doing.

From a business perspective, this is an irrelevant distinction. The requirement was satisfied in a different way, i.e. the engineers satisfying the requirement were replaced by GPT-3, the tool which satisfies the requirement.

I think everyone understood that.

The thread is not about business perspective, it's about the hype around what GPT-3 is and is not able to do.

One thing GPT-3 is not able to do for example, is replacing 2 ML engineers to build a GPT-3 like model. But OpenAI can do that.

It’s not clear what you are trying to say.

Why? I can replace an excellent furniture designer with a much cheaper off the shelf desk.

“AI” replacing the jobs of AI engineers. But we were told it was only going to do that to blue collar work!

Better IDEs have saved countless hours. Saving hours is equivalent to replacing jobs, unless demand is elastic enough to fill that time. Most of the time we are lucky enough that demand at a given price point is much larger than supply, but this won't last forever.

that's not how it works... as it is not a zero sum (i.e. the work is not bounded/fixed but it increases).

IDEs and higher level tools help engineers become more productive. They can do more, with less. This raises the bar on products, and the demand of customers for them (things are pretier, easier to use, etc..), which in turn creates more domains for software to be used, and more demand for engineers.

Google "Induced Demand"

Because most "AI engineering" has lost its meaning and is actually data analysis.

Did it ever have a single meaning? Every company I've been through had a different definition of what "AI engineering" should be

AI engineering: a means of extraction of venture capital from gullible investors :)

Mostly sarcasm, mostly.

We need UBI yesterday.

I would be interested in hearing more about this, within the bounds of what you can share publicly. Most of the touted GPT-3 use cases I've seen to date have dried up or are still in limbo, so hearing about a real production use would be exciting!

Extremely complex classification task is all I can say

GPT-3 is a generative model, isn't it? Can you explain how you converted GPT-3 to a classification model?

There are a couple of ways to do it. You can give it a prompt that shows examples of the classification and it mimics what it thinks is the correct behavior when you feed it new unclassified input. They also have a search endpoint that lets you do classification by giving it an input along with labels as the searchable documents and using the resulting semantic relevance scores.

You can add new "heads" to GPT networks and train those heads to use GPT for new applications.

Not with GPT-3. I believe only Microsoft is allowed to do that.

In general GPT3 is not SotA on (any?) classification task, did you just not have enough data to fine tune a discriminative transformer model? Inference should be cheaper with a smaller transformer/also less lock-in.

I can't go into too much detail here about why we couldn't do that, but one aspect that we found VERY useful is that GPT-3 could draw on real world knowledge not present in the dataset to enhance the results.

Yep few shot learning is a game changer. You don't get perfect results but you can prototype all kinds of systems extremely fast.

Were you a beta user, or is this now open for public access?