[edit]

The current link for the Gatsby unit is wrong. Tried to replace with hyperlink to its homepage http://www.gatsby.ucl.ac.uk but defeated by syntax. May be someone can fix the error? Bill — Preceding unsigned comment added by W102102 (talkcontribs) 11:43, 14 August 2016 (UTC)[reply]

Next acquisition: NHS patient data

[edit]

https://www.newscientist.com/article/2086454-revealed-google-ai-has-access-to-huge-haul-of-nhs-patient-data --91.10.33.46 (talk) 21:14, 5 May 2016 (UTC)[reply]

Requested move 12 September 2016

[edit]
The following is a closed discussion of a requested move. Please do not modify it. Subsequent comments should be made in a new section on the talk page. Editors desiring to contest the closing decision should consider a move review. No further edits should be made to this section.

The result of the move request was: moved as uncontroversial. SSTflyer 05:05, 19 September 2016 (UTC)[reply]



Google DeepMindDeepMindDeepMind – Name of the company has changed, see website. Now operating under the Alphabet holding company, Google branding has been dropped. – Acopyeditor (talk) 22:39, 11 September 2016 (UTC)[reply]

This is a contested technical request (permalink). Anthony Appleyard (talk) 05:35, 12 September 2016 (UTC)[reply]

The above discussion is preserved as an archive of a requested move. Please do not modify it. Subsequent comments should be made in a new section on this talk page or in a move review. No further edits should be made to this section.

Company?

[edit]

Is it really a "company", or was it really started as one? Most of the work they do on WaveNet is open source, and their website reads more like they're a non-profit research institute. I hardly know of any company whose research department is as fully transparent and accessible to the public and their results and working materials all published as open source. --2003:EF:13C6:FE22:9877:F5AE:80F4:F7F6 (talk) 11:47, 1 May 2019 (UTC)[reply]

AlphaCode - comment from Scott Aaronson / the blog Shtetl-Optimized

[edit]

AlphaCode - comment from Scott Aaronson / the blog Shtetl-Optimized

"AlphaCode as a dog speaking mediocre English"

https://scottaaronson.blog/?p=6288

somebody might want to incorporate this into the article

- 2804:14D:5C59:8693:C90B:569:3EFF:444A (talk) 23:01, 6 February 2022 (UTC)[reply]

Request for removal

[edit]

Hi,

I'd like to request that the following be removed from the History section:

In December 2019, co-founder Suleyman announced he would be leaving DeepMind to join Google, working in a policy role.[1]In 2021, the Wall Street Journal revealed that Suleyman had been placed on leave at DeepMind in 2019, following an investigation into employees' allegations that he had bullied them.[2] The company hired an external lawyer to investigate allegations that Suleyman had bullied employees, and he was placed on a leave of absence, before leaving to join Google. An email sent to staff after the story broke, published by Business Insider, said Suleyman's "management style fell short" of expected standards.[3] Business Insider also published further details of historical allegations of bullying against Suleyman, including claims that he boasted about "crushing people", "had a habit of flying off the handle out of nowhere", and demanded employees carry out tasks unrelated to their jobs.[4]

This information is already included in the article focused on Mustafa Suleyman, and is both redundant and irrelevant for inclusion in the DeepMind article, which focuses on the history and information surrounding the company itself.

Bringing this to the attention of Andrevan, who reviewed the edit requests at Talk:Mustafa Suleyman and is therefore familiar with the subject and context.

Thanks for your time, MforSuleyman (talk) 13:47, 27 October 2022 (UTC)[reply]

References

  1. ^ Madhumita Murgia, "DeepMind co-founder leaves for policy role at Google", Financial Times, 5 December 2019
  2. ^ "Artificial Intelligence Will Define Google's Future. For Now, It's a Management Challenge". Wall Street Journal. 26 January 2021.
  3. ^ "Read the leaked email that Google's AI firm DeepMind sent to staff after the bombshell news that its cofounder faced a bullying investigation". Business Insider.
  4. ^ "DeepMind's cofounder was placed on leave after employees complained about bullying and humiliation for years. Then Google made him a VP". Business Insider.

MforSuleyman (talk) 13:47, 27 October 2022 (UTC)[reply]

Sure I'll take a look at this later today or within a couple days Andre🚐 16:21, 27 October 2022 (UTC)[reply]
 Done Andre🚐 19:35, 30 October 2022 (UTC)[reply]
Andrevan, thanks for your help! MforSuleyman (talk) 13:01, 1 November 2022 (UTC)[reply]

Divison of a division?

[edit]

In the article, GOogle Brain is described as a division of Google AI. But Google AI is itself a division (see Google AI)... Am I missing something? 103.114.126.34 (talk) 08:21, 23 May 2023 (UTC)[reply]

You are correct. Google Brain was a division of Google AI, which was a division of Google LLC, which is a subsidiary of Alphabet Inc. InfiniteNexus (talk) 18:11, 23 May 2023 (UTC)[reply]

Deepmind Technologies Limited or Google Deepmind? (2024)

[edit]

I changed the name in the first line to Google Deepmind because that is what is written on the division's website and also that is the title of the article, but User:Ptrnext changed it back to Deepmind Technologies Limited, leaving this comment : "restore legal name" and adding this hidden comment in the source : "full legal name per MOS:FIRSCORP". I don't see anything called MOS:FIRSCORP. And why do we have to use the full legal name (if it indeed is the legal name) only instead of the tradename Google Deepmind, which is also the article title? Or maybe we can keep both, assuming there is a legal name? - Zaheen (talk) 18:36, 12 October 2024 (UTC)[reply]

I meant to write MOS:FIRSTCORP, which requires us to begin company pages with its full legal name. Yes, we can keep both the legal and trade names, like in SpaceX, but usually skip them for the obvious ones (Apple Inc. [legal name] / Apple [trade name]). This edit seems to have removed the citation that was added by another user to validate the legal name. Ptrnext (talk) 19:23, 12 October 2024 (UTC)[reply]
Ptrnext, thanks for the explanation. I have since looked into it and it is indeed legally registered in the UK under that legal name. --Zaheen (talk) 19:56, 14 October 2024 (UTC)[reply]
Still feels confusing that the first sentence says "formerly Google DeepMind" even though it's how the company is currently called. I think the first sentence needs to be changed. Alenoach (talk) 17:49, 15 December 2024 (UTC)[reply]
@Alenoach I've reworded this slightly. It should still be in the same order per MOS:FIRSTCORP. Strugglehouse (talk) 21:55, 15 December 2024 (UTC)[reply]
Mentioning "DeepMind Technologies Limited" first is indeed in line with MOS:FIRSTCORP. But that policy doesn't seem to mention trading names. In my opinion, a sentence like this would be simpler and more helpful for readers:
DeepMind Technologies Limited, known as Google DeepMind (formerly DeepMind), is ...
or a shorter version:
DeepMind Technologies Limited, known as Google DeepMind, is ... Alenoach (talk) 22:07, 15 December 2024 (UTC)[reply]
@Alenoach Both Google DeepMind and DeepMind seem to be used interchangeably. DeepMind is mostly used on the website, but their social media uses Google DeepMind. I have changed the sentence to read:
DeepMind Technologies Limited, trading as Google DeepMind or simply DeepMind, is ... Strugglehouse (talk) 07:52, 17 December 2024 (UTC)[reply]
The company has been renamed from "DeepMind" to "Google DeepMind" in 2023 after the merger with Google Brain.[1] So I believe that Google DeepMind is the standard way to call it now, although some call it "DeepMind" for brevity or by habit. Your new version is better. Alenoach (talk) 20:03, 17 December 2024 (UTC)[reply]

Correcting inaccuracies regarding AlphaChip

[edit]

Hi, I represent Google DeepMind and am requesting edits to provide additional context regarding AlphaChip benchmarks and data availability, including the outcome of Nature's investigation. Requested changes:

Regarding benchmark availability and Nature investigation

[edit]

Current text: "Multiple independent researchers remained unconvinced, citing a lack of direct public benchmarks and independent proof of its claimed superiority over existing commercial chip design tools. Communications of the ACM noted that despite substantial publicity, DeepMind had not provided the comparative benchmarks long requested by experts, leaving some skepticism in the field. Similarly, New Scientist reported that while Google claims AlphaChip has produced "superhuman" chip layouts now used in production, external specialists called for transparent performance data to substantiate these assertions and enable fair comparisons with current state-of-the-art methods." Proposed addition (new paragraph after the above): "The research team released open-source implementation code and training data,[1] and published benchmark comparisons in the peer-reviewed Nature paper[2] and related technical documentation.[3] Following an 18-month investigation into the research's validity, Nature published an addendum in September 2024 that upheld the paper's claims and methodology.[4]" Rationale: This provides readers with complete information about subsequent developments, including independent peer review validation through Nature's investigation. The citations allow readers to access the published benchmarks and data directly.

References

  1. ^ . Google Research Training https://github.com/google-research/circuit_training%7Ctitle=Circuit Training. Retrieved [current date]. {{cite web}}: Check |url= value (help); Check date values in: |access-date= (help); Missing or empty |title= (help)
  2. ^ Nature. 2021. doi:10.1038/s41586-021-03544-w graph placement methodology for fast chip design https://www.nature.com/articles/s41586-021-03544-w%7Ctitle=A graph placement methodology for fast chip design. {{cite journal}}: Check |url= value (help); Missing or empty |title= (help)
  3. ^ Mirhoseini, Azalia (2021). "A graph placement methodology for fast chip design". arXiv:2109.02587.{{cite arXiv}}: CS1 maint: missing class (link) A bot will complete this citation soon. Click here to jump the queue
  4. ^ Nature. September 2024. doi:[DOI if available] title from Nature addendum%5d https://www.nature.com/articles/s41586-024-08032-5%7Ctitle=[Exact title from Nature addendum]. {{cite journal}}: Check |doi= value (help); Check |url= value (help); Missing or empty |title= (help)

Ag77777 (talk) 18:24, 15 October 2025 (UTC)[reply]

Thank you for the feedback. I've corrected the citation formatting below:

Hi, I represent Anna Goldie and AlphaChip and am requesting edits to provide additional context regarding AlphaChip benchmarks and data availability, including the outcome of Nature's investigation.

Requested changes:

Regarding benchmark availability and Nature investigation

[edit]

Current text: "Multiple independent researchers remained unconvinced, citing a lack of direct public benchmarks and independent proof of its claimed superiority over existing commercial chip design tools. Communications of the ACM noted that despite substantial publicity, DeepMind had not provided the comparative benchmarks long requested by experts, leaving some skepticism in the field. Similarly, New Scientist reported that while Google claims AlphaChip has produced "superhuman" chip layouts now used in production, external specialists called for transparent performance data to substantiate these assertions and enable fair comparisons with current state-of-the-art methods."

Proposed addition (new paragraph after the above): "The research team released open-source implementation code and training data,[1] and published benchmark comparisons in the peer-reviewed Nature paper[2] and related technical documentation.[3] Following an 18-month investigation into the research's validity, Nature published an addendum in September 2024 that upheld the paper's claims and methodology.[4]"

Rationale: This provides readers with complete information about subsequent developments, including independent peer review validation through Nature's investigation. The citations allow readers to access the published benchmarks and data directly.

@[X2Step] - I've corrected the citation formatting. All links should now work properly.

References

  1. ^ "Circuit Training". Google Research. Retrieved 30 October 2025.
  2. ^ "A graph placement methodology for fast chip design". Nature. 2021. doi:10.1038/s41586-021-03544-w.
  3. ^ Mirhoseini, Azalia (2021). "A graph placement methodology for fast chip design". arXiv:2109.02587.{{cite arXiv}}: CS1 maint: missing class (link) A bot will complete this citation soon. Click here to jump the queue
  4. ^ "Publisher Correction: A graph placement methodology for fast chip design". Nature. 2024. doi:10.1038/s41586-024-08032-5.

— Preceding unsigned comment added by Ag77777 (talkcontribs) 15:46, 30 October 2025 (UTC)[reply]

 Not done: It's not clear what changes you want to be made. Please mention the specific changes in a "change X to Y" format and provide a reliable source if appropriate. NotJamestack (talk) 21:56, 2 November 2025 (UTC)[reply]
Hi, I represent Anna Goldie and AlphaChip and am requesting edits to provide additional context regarding AlphaChip benchmarks and data availability, including the outcome of Nature's investigation.
Specific changes requested:
Change: After the existing paragraph that ends with "...enable fair comparisons with current state-of-the-art methods."
To: Add the following new paragraph:
"The research team released open-source implementation code and training data, and published benchmark comparisons in the peer-reviewed Nature paper and related technical documentation. Following an 18-month investigation into the research's validity, Nature published an addendum in September 2024 that upheld the paper's claims and methodology."
Rationale: This addition provides readers with complete information about subsequent developments following the criticism, including the release of open-source code addressing reproducibility concerns and Nature's independent investigation outcome.
@NotJamestack Ag77777 (talk) 19:40, 24 November 2025 (UTC)[reply]

Weather model

[edit]

NHC has been using Google DeepMind model for some hurricane forecasts. Don't know exactly how to word any mentions. Source: arstechnica.com/science/2025/11/googles-new-weather-model-impressed-during-its-first-hurricane-season/ Hurricane Clyde 🌀my talk page! 19:27, 5 November 2025 (UTC)[reply]

I added a paragraph on weather prediction to the article. Hope it looks good to you. Frequent Tryer (talk) 02:21, 10 November 2025 (UTC)[reply]

COI edit request by User:Ag77777

[edit]
Thank you for the feedback. I'll clarify the specific changes requested below in the requested format.

Hi, I represent Anna Goldie and AlphaChip and am requesting edits to provide additional context regarding AlphaChip benchmarks and data availability, including the outcome of Nature's investigation.

Specific changes requested:

Change: After the existing paragraph that ends with "...enable fair comparisons with current state-of-the-art methods."

To: Add the following new paragraph:

"The research team released open-source implementation code and training data,[1] and published benchmark comparisons in the peer-reviewed Nature paper[2] and related technical documentation.[3] Following an 18-month investigation into the research's validity, Nature published an addendum in September 2024 that upheld the paper's claims and methodology.[4]"

Rationale: This addition provides readers with complete information about subsequent developments following the criticism, including the release of open-source code addressing reproducibility concerns and Nature's independent investigation outcome.

@NotJamestack Ag77777 (talk) 20:08, 17 November 2025 (UTC)[reply]

 Not done : Duplicated request, I consider the request in the section "Correcting inaccuracies regarding AlphaChip" to be the one that will be analyzed. no longer duplicated Alenoach (talk) 01:30, 25 November 2025 (UTC)[reply]
@Alenoach: After correcting the reply sequence of the edits since October 15 (see other section) and moving this edit request (which was previously inserted in an unrelated section) to the bottom under its own heading, it seems to me like it would be preferable to move forward with this request. The proposed change and text to be added is always the same, but here the references are both included and formatted well, and this section isn't as messy as the other one (so it could be a clean start). If you agree, could you reopen this edit request and strike your previous response? Felida97 (talk) 06:43, 29 November 2025 (UTC)[reply]
Done. It would be good for efficiency if the person who reviews this request also reviews those in Replication crisis, Criticism of Google and Google Brain. Alenoach (talk) 09:57, 29 November 2025 (UTC)[reply]

References

  1. ^ "Circuit Training". Google Research. Retrieved 2 November 2025.
  2. ^ "A graph placement methodology for fast chip design". Nature. 2021. doi:10.1038/s41586-021-03544-w.
  3. ^ Mirhoseini, Azalia (2021). "A graph placement methodology for fast chip design". arXiv:2109.02587.{{cite arXiv}}: CS1 maint: missing class (link) A bot will complete this citation soon. Click here to jump the queue
  4. ^ "Publisher Correction: A graph placement methodology for fast chip design". Nature. 2024. doi:10.1038/s41586-024-08032-5.