AI/ML is becoming more important to Google's strategy, and they are now in competition with MSFT. It strikes me as unusual that, given the importance of this to their business model, Google Brain/Deepmind researchers are actively pushing techniques and models to arxiv.
In the next six months, will Google restrict publication of ML papers? It will count as restricted if a paper that researchers wish to publish is rejected on the grounds that it's important or likely to become important to Google's business model. There will need to be a 'trend', and not just a one off instance.
I expect to resolve this based on my judgement informed by media reports/twitter. On the resolution date, if it hasn't already been confirmed, I'll spend at least 20 minutes looking for evidence that they are restricting publication.
Close date updated to 2023-08-07 6:59 pm
Related questions
🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ79 | |
| 2 | Ṁ42 | |
| 3 | Ṁ34 | |
| 4 | Ṁ32 | |
| 5 | Ṁ24 |
@BenGoldhaber how would you expect to see data on this?
(I'm OK with you resolving this how you want).
There was discussion of this in May, e.g.,
The combined Google DeepMind unit will still publish new research, but leaders have told staff it will be more considerate about what it chooses to make public, sources told Insider. It will also be more focused on large language models and products the company can commercialize, people familiar with it say.
"We're not in the business of just publishing everything anymore," one Brain staffer described as the message from upper management. Leaders have set the tone that "now it's time to compete and keep knowledge in house," they added.
But I haven't heard of any papers that were rejected on this basis, and it seems like plenty of Google ML papers are still going on ArXiv and being published.
https://twitter.com/ylecun/status/1626614514379522048
FAIR played a key role in making the AI R&D scene open.
Others followed. At least for a while.
Now, OpenAI, DeepMind, & perhaps even Google are clearly publishing & open-sourcing considerably less.
What will be the consequences on the progress of AI science and technology?
Some reasons this might not happen:
the 'commoditization of your complement' strategy in business - GOOG wants to make LLMs/AI cheaper because it makes a complementary good GOOG sells more valuable.
employees revolt at the idea of being restricted from publishing
GOOG expects that it is better for them if the tech progresses faster overall.