The Big Story
The Tea on OpenAI
Former OpenAI board member Helen Toner has blessed us with some major (and very concerning) details around Sam Altman’s ousting from the company last November. In a recent podcast, Helen Toner gave us an insider view of the events leading up to the decision to fire Altman followed by his sudden reinstatement.
Outside of OpenAI, Toner has studied artificial intelligence governance and strategy for almost a decade and has worked with policymakers in DC on all things AI. She’s currently the Director of Strategy at the Center for Security in Emerging Technology at Georgetown. As of recently, she delivered a talk at TED2024 in Vancouver discussing the importance of developing smart AI policy, even in the face of significant uncertainty in the technology’s future. Although AI continues to evolve rapidly before our eyes, Toner emphasizes that our incomplete understanding of the tech shouldn’t prevent us from setting effective governance and regulations now. She was warmly welcomed to the OpenAI board in 2021 to help the company in their mission of “building AI for the benefit of all.”
Last November, the board voted to remove CEO Altman after deeming him “not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities.” Toner explains this further during the podcast saying, “The board is a nonprofit board that was set up explicitly for the purpose of making sure that the company’s public good mission was primary, was coming first – over profits, investor interests, and other things. But for years, Sam had made it really difficult for the board to actually do that job by withholding information, misrepresenting things that were happening at the company, in some cases outright lying.”
Toner continued providing specific instances of Altman’s uncooperative behavior, like his failure to inform the board ahead of time of ChatGPT’s launch in November 2022. Instead, board members found out about the human-like chatbot’s debut through Twitter. According to Toner, Altman also failed to mention to board members that he owned the OpenAI startup fund although “constantly claiming to be an independent board member with no financial interest in the company.” Toner went on to reveal that Altman would give the board “inaccurate information about the small number of formal safety processes that the company did have in place.” She and the three other board members eventually grew tired of Altman’s behavior and felt that his breach of trust made it impossible for them to work with him, fueling their decision to fire him.
Following the firing, OpenAI employees rallied and threatened to quit unless the board resigned, and Altman was reinstated. The board also received pressure from Microsoft, which offered Altman a role to lead their new advanced AI research team. After a few chaotic days, Altman was brought back to his executive role, causing the departure of Toner and other board members involved in his short-lived demise. Around this time, rumors surfaced over differences between Altman and Toner over a research paper that she co-penned, where OpenAI’s safety practices were compared to main competitor, Anthropic.
Since the chaos, it’s been business as usual for the OpenAI team, releasing GPT4o and announcing plans to train a new AI model. Earlier this month, they also dissolved their team focused on the long-term risks and safety of AI following its leaders’ departures. Led by Co-Founder Ilya Sutskever and Jan Leike, the Superalignment team was tasked with learning how to “steer and control AI systems much smarter than us.” Sutskever served as Chief Scientist and played a major role in the GPT technology’s development. He also happened to be a proponent of the Altman ouster last year, however ended up taking to Twitter to express his regret for his decision only days later. Leike, a longtime AI researcher, explained his tough decision to leave the company in a lengthy Twitter thread, ultimately saying that the company’s “safety culture and processes have taken a backseat to shiny products.” Altman has since recreated the safety team, which includes him and two other board members.
The departure of OpenAI’s major safety proponents has definitely caused some concern and continues to emphasize the need to set regulations for this powerful technology, especially as we get closer to reaching artificial generative intelligence. It is both an exciting and frightening time as a consumer witnessing the evolution of AI and concurrently watching our future unfold. My advice to all would be to keep an eye and an ear to the developments in AI and to be extra nice to ChatGPT.
Social Updates
To no surprise, TikTok sued the US government earlier this month over the new law signed in place by Biden in April. In a court filing, TikTok states that Congress has “taken the unprecedented step of expressly singling out and banning TikTok,” and called the decision “unconstitutional.” In the complaint, TikTok also claims that the US government has not provided substantial evidence of the Chinese government’s misuse of TikTok. Furthering their point, “TikTok argues that a ban in the US wouldn’t be feasible, as it would force TikTok to move ‘millions of lines’ of software code from ByteDance to a new owner. It adds that limitations from the Chinese government would not allow the sale of TikTok with its algorithm. TikTok claims the ban would make the US version of its app an ‘island’ that gives Americans a ‘detached experience’ from the rest of its users while undermining its business.” Even with the ban looming, TikTok continues to trudge forward, releasing new features and AI powered tools.
According to employees, X, previously known as Twitter, is taking away the ability to see what posts other users have liked to allow for “edgy engagement”. Users will no longer be able to see the likes associated with other people’s accounts or browse someone’s likes through a tab on their profile. The company’s Director of Engineering, Haofei Wang, shared details in his post on X, and notes “the more posts you like, the better your For you algorithm will become.”
Digital Updates
Google Unveils New AI Ad Tools for Search
At the Google Marketing Live conference, Google introduced the new AI-powered features to enhance search and ad capabilities, aiming to help advertisers create more engaging ads. Some of the key updates include testing ads within Search Generative Experience (SGE), which would allow ads to appear directly in AI-summarized search results. Well, that didn’t take long! In addition, Google also launched a tool for creating animations from still images in Product Studio and will watermark all AI generated assets. In the Performance Max platform, advertisers will be able to apply brand-specific fonts and colors in ads. These advancements reflect Google’s push to integrate generative AI across its services amidst the AI race with Meta, Adobe and Microsoft.
Netflix and NFL Sign a Three-Season Deal
For the first time in its history, Netflix will be partnering with a major sports league. Netflix and the National Football League just agreed to a three-season deal that will include showing two Christmas Day games on its service this year. Each year, live sports are a highly sought-after event for advertisers who are looking to tap into a guaranteed audience in real time. This is especially true for the NFL, which has consistently dominated ratings year after year. This move follows Netflix’s push into live events, like the Love is Blind Live Reunion and the recent Roast of Tom Brady. Netflix Chief Content Officer, Bela Bejaria stated, “Last year, we decided to take a big bet on live – tapping into massive fandoms across comedy, reality TV, sports and more. There are no live annual events, sports, or otherwise, that compare with the audiences NFL football attracts.”
A recent alleged leak of internal Google documents has provided unprecedented insight into the company’s search ranking algorithms. Key findings include the importance of link diversity, successful clicks, and the continuing relevance of PageRank. Google also uses historical page versions, but only the last 20 changes for analysis. Brand recognition and authorship significantly influence rankings, while Chrome data and whitelists for specific domains are also utilized. Content freshness, page titles, and document structure are critical ranking factors. This revelation offers SEOs invaluable information to refine their strategies and improve search visibility.
