News: 1679655606

  ARM Give a man a fire and he's warm for a day, but set fire to him and he's warm for the rest of his life (Terry Pratchett, Jingo)

ChatGPT, how did you get here? It was a long journey through open source AI

(2023/03/24)


Opinion When OpenAI released ChatGPT 3.5 in late November 2022, no one expected much from the new release. It was just a "research preview," explained Sandhini Agarwal, an AI Policy researcher at OpenAI. "We didn't want to oversell it as a big fundamental advance," added Liam Fedus, a scientist at the org.

Ha! That was then. This is now.

Unless you've been living under a rock, you know [1]ChatGPT has since become the hottest technology development this decade, heck, maybe this century. At least Bill Gates – you remember him, right? – thinks it's the [2]biggest thing since he was introduced to the idea of a graphical user interface (GUI) in 1980. That led to a product called Windows.

[3]

Amusingly enough, there was nothing all that new in ChatGPT 3.5. It used the same [4]large language model (LLM) as earlier versions. The key difference is that you could now more easily ask questions in natural language instead of accessing it via application programming interfaces (API) or API-driven programs.

[5]

[6]

By making it easy to access ChatGPT, OpenAI, to its surprise, saw it become wildly popular. And, oh, by the way, since [7]Microsoft then invested $10 billion into the business , it seems to have done just fine by the company.

So, great news for open source, too, right? I mean, the company's name is OpenAI, yes? Yes, the name still has open in it, but the source code and the services based on it haven't been open for some time.

[8]

While Google's newly released answer to ChatGPT, [9]Bard , "thinks" that "the GPT-4 model and ChatGPT are both open source projects," it's wrong.

It was meant to be open source, said one of the company's co-founders, another guy you may have heard of named Elon Musk. Musk noted: " [10]OpenAI was created as an open source (which is why I named it "Open" AI), non-profit company to serve as a counterweight to Google, but now it has become a closed source, maximum-profit company effectively controlled by Microsoft. Not what I intended at all."

What happened was Musk first left OpenAI, which was then a non-profit corporation, in 2018, to focus on SpaceX and Tesla. The next year, seeing it would need more money, [11]OpenAI became, for all intents and purposes, a for-profit company . As [12]Sam Altman , OpenAI's CEO, subsequently tweeted, " [13]we will have to monetize it somehow at some point; the compute costs are eye-watering. " I guess $10 billion only goes so far.

[14]

In other words, yet another company failed to figure out how to monetize its open source work. Then, having used open source to build up to GPT-2, it closed the doors on the code.

Besides, even before Microsoft invested big money in OpenAI, [15]Microsoft had exclusively licensed the GPT-3 language model in 2020 .

Mind you, ChatGPT still uses open source code. Just this week when it became clear people could look at others' search history, Altman blamed an open source library. "We had a [16]significant issue in ChatGPT due to a bug in an open source library , for which a fix has now been released and we have just finished validating."

What was the bug? What was the library? We don't know, and I doubt OpenAI will tell us.

Now that ChatGPT is in for the money, it is no longer living up to the Open in its name. As Ben Schmidt, Nomic AI's VP of information design tweeted: " [17]I think we can call it shut on 'Open' AI : the 98 page paper introducing GPT-4 proudly declares that they're disclosing *nothing* about the contents of their training set."

Before all this, you can trace ChatGPT's journey to the open source programs at the heart of AI, machine learning, natural language processing, and deep learning frameworks.

In particular, [18]TensorFlow and [19]PyTorch , developed by Google and Facebook, respectively, fueled ChatGPT. These frameworks provide essential tools and libraries for building and training deep learning models. Without them, there is no ChatGPT.

Another vital open source part of ChatGPT is the oddly named [20]Hugging Face's Transformer . This is the leading open source library for building state-of-the-art machine learning models. It provides pre-trained models, architectures, and tools for natural language processing tasks, allowing developers to build upon existing models and fine-tune them for specific use cases. ChatGPT benefited immensely from the library's support for the GPT series of models, enabling the rapid deployment and scaling of the model.

[21]Bogus ChatGPT extension steals Facebook cookies

[22]GitHub Copilot learns new tricks, adopts this year's model

[23]No reliable way to detect AI-generated text, boffins sigh

[24]Hospital to test AI 'copilot' for doctors that jots notes on patient care

You can see all this in [25]OpenAI's GPT-2 , ChatGPT's direct predecessor. While it got no headlines, GPT-2's impressive capabilities were a result of combining advancements in deep learning, unsupervised learning, and transformer architecture. The open source community played an essential role in the development, testing, and improvement of GPT-2.

How could OpenAI do this? Easily. The open source licenses in the above, TensorFlow, Apache 2.0; PyTorch, modified BSD; and [26]Hugging Face Transformer, support a variety of open-source licenses , including the BSDs. In other words, OpenAI is legally in the clear.

So, like many other programs ChatGPT owes a great debt to open source, but it's not open source, nor is it likely to ever be open source. There you have it. Another depressing open source tale. ®

Get our [27]Tech Resources



[1] https://www.theregister.com/2022/12/03/in_brief_ai/

[2] https://www.gatesnotes.com/The-Age-of-AI-Has-Begun

[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2ZB3XNQEBPz0mx8bMt0d9uAAAAEk&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0

[4] https://writings.stephenwolfram.com/2023/02/what-is-chatgpt-doing-and-why-does-it-work/

[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZB3XNQEBPz0mx8bMt0d9uAAAAEk&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0

[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZB3XNQEBPz0mx8bMt0d9uAAAAEk&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0

[7] https://www.nytimes.com/2023/01/23/business/microsoft-chatgpt-artificial-intelligence.html

[8] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZB3XNQEBPz0mx8bMt0d9uAAAAEk&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0

[9] https://www.theregister.com/2023/03/21/google_bard_ai/

[10] https://twitter.com/elonmusk/status/1626516035863212034

[11] https://openai.com/blog/openai-lp

[12] https://blog.samaltman.com/

[13] https://twitter.com/sama/status/1599669571795185665

[14] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZB3XNQEBPz0mx8bMt0d9uAAAAEk&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0

[15] https://blogs.microsoft.com/blog/2020/09/22/microsoft-teams-up-with-openai-to-exclusively-license-gpt-3-language-model/

[16] https://twitter.com/sama/status/1638635717462200320

[17] https://twitter.com/benmschmidt/status/1635692487258800128

[18] https://www.tensorflow.org/

[19] https://pytorch.org/

[20] https://huggingface.co/docs/transformers

[21] https://www.theregister.com/2023/03/23/chatgpt_fake_chrome_extension/

[22] https://www.theregister.com/2023/03/22/github_copilot_learns_new_tricks/

[23] https://www.theregister.com/2023/03/21/detecting_ai_generated_text/

[24] https://www.theregister.com/2023/03/21/abridge_hospital_ai_doctor/

[25] https://github.com/openai/gpt-2

[26] https://huggingface.co/docs/hub/repositories-licenses

[27] https://whitepapers.theregister.com/



Open or Closed...

Great Bu

I don't mind if it is open or closed source as long as it keeps doing my homework for me.......

Re: Open or Closed...

Sp1z

Good luck when you get to the exam........

Re: Open or Closed...

Great Bu

The only exams I get these days are on my prostate.....

Re: Open or Closed...

Anonymous Coward

I hope you passed, or at least can pass when you need to.

Old guy.

unimaginative

It sound as if they are doing somerhibg wrong, but the entire point of using a BSD/MIT type license (rather than a GPL type license) is to allow propreitary redistribution.

Yes, it would be great if it was opensource, but they are not doibg amything sneaky by keeping it proprieatryc

zuckzuckgo

Other organizations are still free to assemble the same open source components into a competitor, they just can't access whatever "special sauce" was added by OpenAI. And it's probably just mayonnaise.

Americans would say

elsergiovolador

Schmucks - all those developers giving out their time for free, missing out on important life events, family, self-development...

Just to get those few stars on GitHub that nobody cares about or get a few more followers.

Then maybe one day someone from FAANG will notice and schedule the dream interview.

In the meantime big corporations have other ideas.

Make billions and don't pay a penny back, and not a penny forth.

So, NotOpenAI is monetizing ?

Pascal Monett

Vote with your wallet, people.

What about borrowing (stealing) the code

matski

The truth is that OpenAI stole the licensed code and is keeping it inside itself as a part of it's model. They just claim it was "public" so they could take it. This was allowed when it was doing the "research" and non-profit but right now it's a commercial software. And they are using this original code (that's tokenized and stored in some form) to generate answers and code without showing any license or even credit for the original creator. This is a massive legal issue - this product is flawed and products that use the code it produces can be later sued for infringement. It did so for every other piece of work - science papers, art, etc.

Re: What about borrowing (stealing) the code?

Long John Silver

Your remarks illustrate the mess we get into when ideas, and when easily replicable digital code, are made into a pretence of property.

Just look at what already goes on in the area of software copyright and software patent disputes. Consider too, the rotting rambling edifice of 'intellectual property' supposed rights across the board of industry and culture. A framework of 'rights' so extensive, so convoluted, and so prone to ambiguity and contradiction, that bright ordinary people understand little, and trained lawyers need to specialise in particular aspects. All of this arising according to the dictum of a false proposition giving rise to all propositions true and false : the origin of this mess being the false supposition of ideas being equatable to physical property.

The only possible currency for valuing the worth of ideas rests on attribution and praise.

Re: What about borrowing (stealing) the code

MOH

Its not just the code though. The training data too.

The whole thing is essentially a giant plagiarism machine. Sure, it'll chop and change a bit, but ultimately all it's doing is regurgitating other people's content.

Where is Robin Hood?

Long John Silver

Eventually, various incarnations of Robin Hood will release corralled code into the wild. Once gone, as IP rentiers in general are learning, no amount of huffing and puffing, no number of erroneous cries of "Theft!", and no host of avaricious lawyers, can coax the code back into the corral.

Scattered among academic and commercial centres across the globe, the code will be used, extended, and improved. Unless by then, all Western universities have been converted into "for maximum profit" institutions, as demanded by Neo-liberal doctrine, academics shall seek kudos flowing from open publication of their efforts. Then there is the Far East and 'Global South' to consider: already they chaff at Western (mostly USA, but cheered on by 'lesser' nations) restrictions on use of ideas and trade of products and services derived therefrom.

The days are numbered for behemoths such as Microsoft. The pace of change in the 21st century rapidly assigns them to anachronism. Fleet-footed small firms, co-operatives, and cottage industries shall dance between the legs of ailing dinosaurs.

When you're not looking at it, this fortune is written in FORTRAN.