Pages

Wednesday, November 22, 2023

On March 11, 2019, OpenAI issued an online public disclosure statement of a limited partnership contract, something where candor among the authors, and subsequent management officers, each to one another, might matter. A limited partnership was announced, where limited partners might or might not have had fiduciary duties toward the general partner. Particularly as things evolve over time. Ostensibly, the nonprofit's Board holds preeminant authority.

https://openai.com/blog/openai-lp

Read that item. Two individuals are named as authors.  While vague on much detail, there is disclosure that OpenAI Nonprofit is creating (or at that time already had created) a limited partnership OpenAI LP. It was apparent OpenAI Nonprofit was general partner. Use of the term "Investors" might be unclear but the limited partnership aspect is clear:

Our mission is to ensure that artificial general intelligence (AGI) benefits all of humanity, primarily by attempting to build safe AGI and share the benefits with the world.

We’ve experienced firsthand that the most dramatic AI systems use the most computational power in addition to algorithmic innovations, and decided to scale much faster than we’d planned when starting OpenAI. We’ll need to invest billions of dollars in upcoming years into large-scale cloud compute, attracting and retaining talented people, and building AI supercomputers.

We want to increase our ability to raise capital while still serving our mission, and no pre-existing legal structure we know of strikes the right balance. Our solution is to create OpenAI LP as a hybrid of a for-profit and nonprofit—which we are calling a “capped-profit” company.

The fundamental idea of OpenAI LP is that investors and employees can get a capped return if we succeed at our mission, which allows us to raise investment capital and attract employees with startup-like equity. But any returns beyond that amount—and if we are successful, we expect to generate orders of magnitude more value than we’d owe to people who invest in or work at OpenAI LP—are owned by the original OpenAI Nonprofit entity.

Going forward (in this post and elsewhere), “OpenAI” refers to OpenAI LP (which now employs most of our staff), and the original entity is referred to as “OpenAI Nonprofit.”

Who invests how much when, how money from the LP is divied up, are unclear aspects of the announcement. Continuing:

The mission comes first

We’ve designed OpenAI LP to put our overall mission—ensuring the creation and adoption of safe and beneficial AGI—ahead of generating returns for investors.

The mission comes first even with respect to OpenAI LP’s structure. [...]

OpenAI LP’s primary fiduciary obligation is to advance the aims of the OpenAI Charter, and the company is controlled by OpenAI Nonprofit’s board. All investors and employees sign agreements that OpenAI LP’s obligation to the Charter always comes first, even at the expense of some or all of their financial stake.

That referenced Charter says, in part:

Broadly distributed benefits

We commit to use any influence we obtain over AGI’s deployment to ensure it is used for the benefit of all, and to avoid enabling uses of AI or AGI that harm humanity or unduly concentrate power.

Our primary fiduciary duty is to humanity. We anticipate needing to marshal substantial resources to fulfill our mission, but will always diligently act to minimize conflicts of interest among our employees and stakeholders that could compromise broad benefit.

Back to the OpenAI LP announcement:
Purple Notice
Our employee and investor paperwork start with big purple boxes like this. The general partner refers to OpenAI Nonprofit (whose legal name is “OpenAI Inc”); limited partners refers to investors and employees.

Only a minority of board members are allowed to hold financial stakes in the partnership at one time. Furthermore, only board members without such stakes can vote on decisions where the interests of limited partners and OpenAI Nonprofit’s mission may conflict—including any decisions about making payouts to investors and employees.

Those were "ground rules" under which Altman as CEO functioned, with the nonprofit's Board holding ultimate control. "All humanity" language is interesting.

NEXT -

Consider The Verge, at an item updated Nov. 15, days before Altman was fired by the OpenAI nonprofit's board: https://www.theverge.com/23961007/microsoft-ignite-2023-news-ai-announcements-copilot-windows-azure-office   

The Verge publishes a linked Nov. 15 item: https://www.theverge.com/2023/11/15/23960517/microsoft-copilot-bing-chat-rebranding-chatgpt-ai

Both Verge items were about the 2023 Microsoft Ignite dog-and-pony show happenings, in particular news about new things MS would do, or had implemented but not previously disclosed publicly. 

The linked Verge subitem was interestingly titled: "Microsoft rebrands Bing Chat to Copilot, to better compete with ChatGPT / Microsoft has settled on Copilot as its main brand for its AI chatbot, after months of trying to push Bing Chat."

The first linked Verge item had info on a number of things, including:

Microsoft launched its big AI push earlier this year as part of its Bing search engine, integrating a ChatGPT-like interface directly into its search results. Now less than a year later, it’s dropping the Bing Chat branding and moving to Copilot, the new name for the chat interface you might have used in Bing, Microsoft Edge, and Windows 11.

Microsoft initially talked up the Google search competition for its AI ambitions earlier this year, but it now looks like it has its sights set on ChatGPT instead. The Bing Chat rebranding comes just days after OpenAI revealed 100 million people are using ChatGPT on a weekly basis. Despite a close partnership worth billions, Microsoft and OpenAI continue to compete for the same customers seeking out AI assistants, and Microsoft is clearly trying to position Copilot as the option for consumers and businesses.

Read Article >

If you "Read Article" you will see it titled as noted, the article being the second Verge link given above, and it stated:

[,,,] “Bing Chat and Bing Chat Enterprise will now simply become Copilot,” explains Colette Stallbaumer, general manager of Microsoft 365. The official name change comes just a couple of months after Microsoft picked Copilot as its branding for its chatbot inside Windows 11. At the time it wasn’t clear that the Bing Chat branding would fully disappear, but it is today.  [image omitted]

Microsoft is now pitching Copilot as the free version of its AI chatbot, with Copilot for Microsoft 365 (which used to be Microsoft 365 Copilot) as the paid option. The free version of Copilot will still be accessible in Bing and Windows, but it will also have its own dedicated domain over at copilot.microsoft.com — much like ChatGPT.

Business users will sign into Copilot with an Entra ID, while consumers will need a Microsoft Account to access the free Copilot service. Microsoft Copilot is currently officially supported only in Microsoft Edge or Chrome, and on Windows or macOS.

When Microsoft announced Bing Chat earlier this year, the company described the chatbot as an “AI-powered copilot for the web,” and since then we’ve seen the company use the Copilot branding for a number of AI efforts after GitHub originally used the Copilot name last year.

This new rebranding means Copilot is becoming more of a standalone experience that you don’t have to navigate to Bing to access anymore. Bing is simply part of what powers Copilot now. But the move away from Bing is an interesting one, given Microsoft put a lot of effort into launching its AI efforts inside its search engine and positioned it as a way to steal market share from Google.

Microsoft claims Bing is still a big part of Copilot, though. “Bing remains a prominent brand and technology powering many Copilot experiences while continuing to be a leader in the search industry,” says Caitlin Roulston, director of communications at Microsoft, in a statement to The Verge. [again image omitted]

 At the time of the Bing Chat launch earlier this year, Microsoft held an internal Q&A for employees to get answers about its AI search push. Sources familiar with the meeting tell The Verge that Yusuf Mehdi, Microsoft’s consumer chief marketing officer, explained why the company was sticking with Bing at the time instead of a new brand like Microsoft Copilot.

“It’s a neutral vessel, so all the research from the branding team shows that people are basically neutral on Bing, which is generally a good thing,” explained Mehdi in an internal meeting with Microsoft employees in February. He also detailed that awareness of the Bing brand was worth around $200 million dollars. “So we said ‘do we want to start from scratch or build on that?’ It has all the positive things, it’s four letters, it has one syllable, it’s global, and it has equity. So we said we’re gonna stick with the Bing brand.”

Bing will no longer be the main entry point for Microsoft’s AI ambitions with Copilot anymore, and it’s not clear if the push for AI search was ever successful for the software maker. Microsoft CEO Satya Nadella called Google an 800-pound gorilla that he wanted to make dance earlier this year, but Google hasn’t rushed to integrate AI into its search results in quite the same way as Microsoft. And nearly 10 months after the Bing Chat launch, Google is still at over 91 percent market share according to StatCounter.

Update, November 15th 2:45PM ET: Article updated to clarify the rebrand is just to Copilot, although Microsoft routinely refers to Microsoft Copilot as the branding for its collection of AI-powered copilots.

 Where things are most interesting is whether the OpenAI Nonprofit board knew as much about Microsoft's intent to make Copilot a stand alone as Altman might have known. Also, Microsoft owns GitHub, and can use content to feed an LLM to present coding specific AI where OpenAI might have been excluded from any GitHub shared access. Did the board know whether stuff rolled out at OpenAI DevDay had previously been licensed in some way to Microsoft, if that was the case, to where Microsoft could greatly enhance the Copilot effort beyond the Bing Chat initial effort.

BOTTOM LINE: How does the Copilot approach fit with Altman's ongoing duty to keep the board informed of all he knew of Microsoft and OpenAI's dealing with MS as a main source of development capital. There is this:

If the outside board members felt blindsided by things learned at Microsoft Ignite, they may have regarded Altman as not keeping them sufficiently informed of Microsoft's licensed rights to use OpenAI intellectual property. It seems handling of intellectual property rights to stuff revealed at OpenAI DevDay might have had independent board members feeling blindsided. Something of that sort, in days before the board lost confidence in Altman, seems to have been a material factor in the board losing confidence. The board has been silent regarding details, so Crabgrass is clearly guessing. 

A final thought - Does this Nov. 15, 2016 OpenAI post - authored by Brockman, Sutskever and Altman about OpenAI staff working on the Azure platform, mean that the model, the data and training network, the trained final nets, and interface software is all on Azure servers, and not on separate OpenAI owned hardware? 

Might the model be on Microsoft servers, with some sharing arrangement between OpenAI and MS (via Azure) exist, (some MS billions invested possibly being Azure access hours), and was the board fully advised of any ancillary understandings or contract between the two firms re the use of Azure decision, something between Microsoft and OpenAI LP about access/ownership/sharing arrangements on Azure servers? (Under special vs standard Azure cloud purchasing terms?)

In effect, where does the entire serial model set for OpenAI GPT 4 and earlier reside, along with new GPT 5 measures and the DevDay tools, and under what terms and conditions? If there were loose oral understandings, Altman with Microsoft people, then would worry over that, in light of Microsoft Ignite events, have induced board action?

Possibilities abound, but the dismissal followed OpenAI DevDay and MS Ignite.

So, what in there got the board aggressive? Under what hardware and IP arrangements following the two promotional events? Is there a link, or no link?

In effect, who owns what IP - and/or access to it, by special arrangement, if any?

UPDATE: With Nadella having gotten hot over the firing then tweeting of looking forward to working with OpenAI into the future with its new CEO, and with Altman at Microsoft now doing AI work, is there great rancor or is this a case of bedfellows having a spiff where things are now well resolved to where business on both sides can cordially continue? It appears Nadella holds all the trump cards, and MS stock went up to show the market believes Nadella holds the hammer, if needed, but cordial relations might still continue (or be necessary for the best future of both ventures). OpenAI, is it traded to where the market would react, or is it privately held where some IPO was/is  envisioned prior to the firing or now? It does not seem to be a death knell for OpenAI, but how the market might value an IPO in the future depends upon how things between the two ventures evolve. Now is not IPO time.

Future aspects of the nonprofit with a related for profit limited partnership structure which OpenAI evolved might well evidence whether such a structure is a good idea, or otherwise, as a way others may do business in the future. 

The bot could be asked about that, "Give a two paragraph analysis about, etc."

But would the bot hallucinate in answering?

FURTHER UPDATE: If the parties get lawyered up to bury the hatchet, good could come of it; whereas lawyered up to endlessly litigate would likely turn out really ugly. Could things patch up as an "in our mutual interest" thing, without changes in the OpenAI board? Nadella seems to have offered an olive branch via his tweet. And the new CEO, Emmett Shear, presumably would not have accepted the job if all he foresaw was a giant headache. Crabgrass believes in business adults being expected to be rational, or hopes it will be so.

FURTHER UPDATE: With regard to a possible mass walkout at OpenAI, or flyout to Redmond from the Bay Area, some may agree with Sutskever and want to stay and work on his presently driving research concern, prevention of a superintelligence being created and going rogue, while Microsoft and other Seattle metro area tech firms have laid workers off and MS might be hesitant to take on a new 600 -700 career path bodies, sight unseen. Some who say they'd follow Altman might not follow through, and might wonder about Microsoft culture they don't know vs OpenAI culture they know. Altman and Brockman will know who they want to poach and who they'd leave at OpenAI, especially if ongoing collaboration is to happen where they'd want some human intelligence to continue at that other AI venue.