OpenAI’s Agency Complexities: A Reflection on Business Models & Technology
The unexpected OpenAI agency problems are a great opportunity to reflect on Business models and Technology.
By now, everybody has read about the uncommon corporate structure of OpenAI which includes:
(A) a non-profit entity whose mission is `To build safe and beneficial artificial general intelligence for the benefit of humanity `
(B) a for-profit entity governed by the non-profit that has agreed to cap its return on investment to 100x.
According to OpenAI`s website:
`we launched with the goal of $1 billion in donation commitments. Yet over the years, OpenAI’s Nonprofit received approximately $130.5 million in total donations`
What we have been calling investments are actually categorized as philanthropic “donations to the non-profit and governing organization. Cindy Chin (NASA Datanaut and VC whisperer) with experience on both for-profit and not for profit Boards, reports:
`Also of note is ownership structure. Sam Altman does not own direct shares of OpenAI nor does other “investors” which looking at the tax filing documents. They are categorized as philanthropic “donations.” $50M 2018 and $32M in 2019, etc. for OpenAI the non-profit and governing organization. The LP tax documentation on the limited for-profit arm and fund is private.`
What we haven’t been thinking about is who owns the Intellectual property of the Large Language Model developed & continuously being improved by the non-profit entity. Elon Musk spoke about this more than 6 months ago but of course, none of us was paying much attention as we were in ChatGPT Nirvana.
In a nutshell, I am realizing that the rights to the weights of an LLM are effectively the rights of the model’s learned behavior and also the potential for customization. The rights to its inference system means in plain language, the rights to its acquired capabilities from training on huge datasets that allow it to recognize, summarize, translate, predict, and generate text and other forms of content.
OpenAI has been recognized as the leading private company in the next generation of Conversational Chatbots powered by LLMs. As a result, their valuation exploded despite the fact that they weren’t profitable. This is not uncommon at all in the world of innovation.
Sam Altman became the face of OpenAI and GenAI. This is a testament that our fascination with public figures and Human brands is 100% intact. `Humans love human brands` quoting Thomas Power.
OpenAI undoubtedly qualifies as a major disruptor not just as an innovator that offers better, cheaper, faster services/products. At the same time, we need to recognize that OpenAI is a business with a traditional organizational structure and culture. It is a business with a leading AI product that continues to innovate and launch new AI-products and services. They started with the mass consumer AI-offering of ChatGPT, added enterprise-level access to their LLM, and as the Black Swan event hit them, they had just launched the GPT store for developers.
OpenAI is a traditional Digital native business designing all sorts of AI products powered by their evolving LLM. Despite the cap on the shareholder return on investment, their Business Strategy and Culture is to maximize shareholder value. They are a Web2 type of company in which value accrues to the business and the shareholders. OpenAI is not a business that rewards its creators and or users. OpenAI is setup with the incentives of a Traditional Digitally native business.
OpenAI is not an AI-first business and is not a Web3 business. The business itself can be run without using AI. The strategy of the business, the entire value chain of the business is not run or based on AI. OpenAI`s products are great for customers but there is no innovation in terms of incentives for their creators or users.
OpenAI has no element and no intention to move towards a Decentralized Autonomous Organizational Structure (DAO). Its governance and incentive structure are as Traditional as it gets. This is extremely important in my opinion because of the nature of their product and mission — AGI for Humanity.
By now, it should be clear that we must build decentralization in all next-generation AI infrastructure. We cannot rely on Sam, Elon, Satya, and any one rising Human Brand — either a brilliant entrepreneur or a brilliant corporate leader — because the complexity of agency problems is not only unimaginable but dangerous for humanity.
Before you stop reading because you are so opposed to DAOs as unrealistic structures, I invite you to reflect on Decentralization and consider that it comes in many flavors. There are degrees of Decentralization much like there are different forms of Digitalization. For example, a Digital native business is run on the Cloud (that is a basic layer requirement but not sufficient). A Decentralized business is run on some Consensus mechanism (that is a basic layer requirement but not sufficient).
Let’s put all this in perspective in financial services which is one of the industry sectors expected to largely benefit from the adoption of GenAI (e.g., fraud, credit, investing, etc.)
Chris Skinner`s new tagline `you no longer need to think about money because money thinks for you` is not at all far-fetched because of GenAI. Yes, we can and will move to what I call `Intelligent Digital Wallets` or `Intelligent Digital interfaces` that advise us and largely auto-magically optimize our finances. Just this past week, we saw several such innovative interfaces powered by GenAI. At the Singapore Fintech Festival, MYbank the SME-focused Digital Bank showcased LARK, a Multimodal Conversational Interface for the credit needs of SMEs. A great example of designing an advanced AI product for financial inclusion.
However, all these Intelligent Interfaces, don’t change the reality that our bank account can be frozen from our Banking provider, or our Banking provider can and will change fees or stop serving one of our needs because it is unprofitable for them, etc.
GenAI can’t predict or solve challenges with sanctions, capital controls, and new unfavorable regulations. GenAI doesn’t rid us of Digital Feudalism (maybe it will affect which entities are the Feudalists). GenAI will not provide us with a fairer Sovereignty than the current one. GenAI hasn’t been integrated yet into corporate structures, Strategic decision-making, and business models.
While we recognize GenAI Fintech innovators like MYbank (the pioneer of the 3–1–0 credit model [1] ) or AlphaSense (a leader in market and competitive intelligence solutions [2]), these entities do not use AI to run their businesses. They only use AI at best, to discover, design, and launch new products and services.
Don’t be blindsided by shiny AI-products. We are early in the AI development cycle although in a hype, so it is natural to lose our orientation as we shape the next era of the Internet.
All these shiny AI businesses are run in a very traditional way. They are competing to get on the current leaderboard of Digital Feudalism and displace the Magnificent 7.
We must design the next generation of companies as Decentralized AI-first companies not just Digital native companies with AI products and services, run on Traditional corporate structures.
The corporate structures of the companies that are building core infrastructure for an AI-first world, must be more decentralized and should start eating their own lunch by using AI to run their business.
`To build safe and beneficial for humanity AI-first businesses we need to decentralize decision making and embed AI beyond the product/service`.
I invite you to listen and reflect on a timely discussion with Ben Goetzel on the topic of Advanced Technologies should run on decentralized infrastructures
[1] Using AI for digital lending by MYBank https://youtube.com/shorts/h8JTJ2--TIo?si=jvXZ6Rkof4XoUvoe
[2] Read more about AlphaSense in the section `Four Fintechs` of my article `ChatGPT Empowering Investment Research: Trends and Use Cases` https://medium.com/datadriveninvestor/chatgpt-empowering-investment-research-trends-and-use-cases-45e3ee926a92
📌 Twitter: https://twitter.com/efipm
📌 Subscribe to my YouTube Channel with my insights and industry leader interviews. New video every Wednesday: https://www.youtube.com/EfiPylarinou
📌 Apple Podcast 👉 https://buff.ly/3P1Cp1Z
📌 Spotify Podcast 👉 https://buff.ly/3xPyWaV
📌 Linkedin: https://www.linkedin.com/in/efipylarinou/
📌 TikTok: https://www.tiktok.com/@efiglobal
📌 Web: https://efipylarinou.com/