Unpacking The OpenAI Meltdown
Dr. Brian Massey
Managing Partner at Anordea | Strategy Advisor, Professor & Keynote Speaker | Helping CXOs Drive Faster AI Transformations across Banking and Financial Services.
Hello and welcome to the latest edition of AI Strategy Brief. In this edition we’ll review some recent industry developments in AI, and try to make sense of the recent OpenAI meltdown.
?
AI Industry Updates:
?
Technology - Microsoft Launches Copilot 365 & AI Chips
Microsoft has introduced Microsoft 365 Copilot, launched on November 1, 2023 for limited users. The new tool integrates OpenAI's ChatGPT technology into Microsoft Office applications and is priced at $30 per user per month, targeting E3, E5, Business Standard, and Business Premium customers. In addition, Microsoft has integrated Bing Chat into the Microsoft 365 experience. The Enterprise AI market, projected to reach $88 billion by 2030, is a key focus for Microsoft, particularly in light of similar advancements by competitors like Google.
?
Added to this, on November 15, 2023, Microsoft unveiled two significant advancements in AI and cloud computing technology at the Ignite Developer Conference. The Azure Maia 100, a custom cloud computing chip optimised for generative AI tasks, is the first in the Maia series and boasts 105 billion transistors using 5-nanometer process technology. Alongside, Microsoft introduced the Azure Cobalt 100, its first in-house microprocessor for cloud computing, featuring 128 computing cores and a 40% reduction in power consumption compared to other ARM-based chips. These innovations position Microsoft alongside Google and Amazon in offering custom silicon for cloud and AI applications and are expected to support programs like GitHub Copilot and generative AI from OpenAI.
?
Technology - OpenAI Launches Custom GPTs
OpenAI has announced the roll-out of customisable versions of ChatGPT, termed GPTs, designed for specific purposes. These GPTs can be tailored for various tasks, shared with others, and require no coding to build. Users can make GPTs for personal, company internal use, or for public sharing. The process involves starting a conversation, providing instructions and additional knowledge, and choosing features like web search, image creation, or data analysis. OpenAI is introducing a GPT Store later this month, where users can share their GPTs publicly. This store will feature creations by verified builders and categorise GPTs based on their utility, like productivity or education. Creators can also earn revenue based on the usage of their GPTs.
?
OpenAI has implemented systems to review GPTs against usage policies to prevent sharing of harmful content and has introduced a verification process for builders. GPTs are envisioned to evolve into more sophisticated 'agents' over time, capable of performing real-world tasks. Developers can enhance GPTs by connecting them to external APIs, allowing integration with databases, emails, or e-commerce platforms. For enterprise customers, GPTs offer a high level of customisation for specific business use cases. Companies can create internal GPTs for tasks like crafting marketing materials or assisting in customer support.
?
Finance – AI Used To Assess Earnings Calls
Investors are now turning to AI, not just for documents analysis, but also for detecting subtle cues in speech that might reveal true emotions and intentions. AI speech recognition and analysis has recently been used to gauge CEO truthfulness, as in the example of Francis deSouza’s final earnings call at Illumina. This shift towards AI analysis marks a new level in the ongoing strategic interplay between fund managers and company executives. Traditional natural language processing (NLP) techniques focus on the content of speeches, but the new approach adds an audio dimension, detecting elements like hesitations, filler words, and imperceptible "microtremors."
?
Robeco, a large asset manager, has already integrated audio signals into its investment strategies with positive results. However, this approach is not without challenges. It can be expensive and requires robust technological infrastructure. Also, there's a risk of bias in the algorithms, especially related to factors like gender, class, or race. Moreover, analysing speeches from different executives or in different languages can be complex. For executives and leaders, the challenge here is that modulating voice is typically more difficult than choosing words, providing a new frontier in the strategic communication between companies and investors.
?
?Unpacking The OpenAI Meltdown
If you’ve been following the AI space recently, you cannot have missed the drama unfolding at OpenAI over the last week. What happened and why does it matter? Let’s take a look.
?
Key Players:
?
How Did This Start?
The drama began last Friday when the Board of OpenAI voted to remove Sam Altman as CEO with immediate effect. Unable to support this motion, Greg Brockman, President of OpwenAI, resigned with immediate effect in support of Sam. The Board then went ahead and released the news to the world that Sam had been removed as CEO without citing reason or cause.
?
领英推荐
How The Story Developed
If you followed the story over the weekend and into this week, you will have seen the chaos that ensued, with more plot twists than a season of Game of Thrones. For those that have not followed the story, I will try to briefly summarise what we know so far. With Sam and Greg gone, the Board moved to appoint Mira Murati CTO of OpenAI as Interim CEO. Mira’s first act was to try and rehire Sam and Greg, which the Board stopped immediately. Mira then took to social media to support Sam and Greg with the message ‘OpenAI is nothing without its people’. This rallying cry was soon taken up by other members of OpenAI staff on X.
?
Simultaneously, investors including Microsoft and VC firms learned about Sam’s dismissal without any fore warning or consultation, effectively blindsiding 98% of the shareholders in OpenAI LLC. Being deeply unhappy with this dismissal, the investors began working to have Sam reinstated. Microsoft also made an offer to Sam and Greg to set up and run a new AI SBU within Microsoft Group. In an effort to resolve the situation Sam met with the Board on Sunday but they were unable to find common ground with Sam requesting full reinstatement for himself and Greg and a change of the Board.
Monday ended with Satya Nadella, CEO of Microsoft giving an impromptu interview on Bloomberg to reinforce his support for Sam and Greg. Throughout Monday, OpenAI staff continued to show support for Sam and Greg on social media and by signing a petition to have Sam and Greg reinstated and the Board changed.
?
Ultimately almost all OpenAI staff signed this petition stating they had no confidence in the Board. Simultaneously the Board of OpenAI recruited a new Interim CEO by tapping the former CEO of Twitch, Emmet Shear. In a further plot twist, Shear requested that the Board explain to him the reasoning behind Sam’s dismissal, stating he would not take up the role without clear reasoning being communicated to him. In the discussions that continued between all stakeholders, it became clear that Sam and Greg has the backing of investors and all staff, eventually leading to the Board to concede their position and step down.?
Where Are We Now?
As of writing (22/11/23) three of the original Board members have been dismissed and a new Board has been set up including former Salesforce Co-CEO Bret Taylor and former Treasury Secretary Larry Summers. Adam D’Anglo retains his seat on the Board for now. Sam Altman has been reinstated as CEO of OpenAI.
?
What Caused This? – The Poe Hypothesis (Competitive Motives)
One theory on what caused this chaos relates to Adam D’Anglo and his role as CEO of Quora. Recently Quora had created an ecosystem for developers to build custom AI programs, with its Poe application and ecosystem. This gave Quora a leading position in AI Application Customisation and Monetisation. That is until, OpenAI announced their Custom GPT feature and coming GPT Store at the recent OpenAI Developer Day. Some have speculated that the move against Sam was in response to this, and a form of outright competition. Given Adam D’Anglo retaining his seat on the Board, this theory seems less likely, but clearly it is questionable to have competitors sitting on your Board.
?
What Caused This? – The Effective Altruism Hypothesis (Ideological Motives)
Another speculative view relates to the rise of an ideological movement called Effective Altruism in the tech industry and beyond. Several members of the Board, including Ilya, Tasha and Helen are reported to be proponents of this movement, and have serious concerns about the speed of AI innovation for safety and ethics reasons. This stance is in direct opposition to that of ‘Accelerationists’ like Sam and Greg, who are thought to believe in innovating and developing technologies as fast as possible, while trying to maintain safety and ethics throughout the process. Some believe that Sam being at odds with the Board ideologically, especially given recent OpenAI commercialisations, is what led to this event.
?
Why Is This Important?
Aside from providing followers of the AI space with an unexpected drama (David Fincher has already been suggested as a potential director for an OpenAI film), there are very real reasons why people working in AI, and beyond, should care about all this. Firstly, it highlights the problems of misaligned board and organisational missions, which is deeply problematic in a space where effort is being deployed on generating new technologies as impactful as AI. Secondly, it emphasises the fragility of the current AI Application Ecosystem which many companies are starting to build their operational capabilities around. Clearly from this episode, OpenAI was a lot less stable than people may have thought. However, at this moment, OpenAI is still the only real option available to most small and medium sized firms wishing to build on and integrate some form of Generative AI into their operations. Given the advantages these new AI technologies offer, no leader can really afford to ignore them. This would seem to leave companies with a serious problem around building capabilities on uncertain technological foundations, what I call the Layering Problem of AI. If OpenAI had of disappeared this week, where would that leave your AI Strategy?
Leadership Takeaways:
?
That’s it for this edition of AI Strategy Brief. Subscribe now, and if you like what you read today, please like and share it with your network to help me reach a wider audience. Have a great day, and I'll see you next time!