OpenAI's GPT-4o mini ... US/UK investigate Microsoft's hiring Inflection AI staff ... Coalition for Secure AI... TL;DR 22Jul24

Last update: Monday 7/22/24 

Welcome to our 22Jul24 TL;DR summary of the past week's top AI stories on our "Useful AI News" page  ➡ 1) OpenAI's new cheaper, smarter GPT-4o mini, (2) US/UK investigate Microsoft's hiring Inflection AI staff, and (3) Coalition for Secure AI. Where possible, our summaries note connections between this week's events and past events.

                OpenAI Logo

TL;DR link ➡ HERE  

A. TL;DR summary of Top 3 stories 

1. GPT-4o mini | 2. Inflection | 3. AI security

Last week was a relatively slow week for AI news; not much happened. Correction: not much happened that was really new. The most important developments were unsurprising reflections of strong trends that have come to dominate generative AI developments. 

1) OpenAI's cheaper, smarter GPT-4o mini

Large language models have yet to prove themselves to be cost-effective investments for large enterprises. Howeve, we recently learned that OpenAI is making more money from its GenAI products than Microsoft. It seems that small to mid-size customers have found its services to be reliable enough for their limited applications to be cost-effective.

So it came as no surprise that OpenAI had joined its partner Microsoft and its competitors Google and Anthropic in offering a smaller model called GPT-4o mini that was made even more cost-effective because it was cheaper than OpenAI's other models. OpenAI did not specify the size of its mini model, i.e., the number of parameters; so we don't know if its as small as Anthropic's Haiku, Google's Nano, or Microsoft's Phi-3.

BackToTop

2) US/UK investigate Microsoft's hiring Inflection AI staff

Google, Microsoft's strongest GenAI competitor, had purchased DeepMind in 2024. In 2024 Google repositioned DeepMind's founder, Dr. Demis Hasabis, to  head all of Microsoft's GenAI operations in response to the success of OpenAI/Microsoft/s ChatGPT. 

In March 2024, Microsoft hired most of the staff of a startup called InflectionAI, whose CEO, Mustafa Suleyman, was another cofounder of DeepMind; Microsoft compensated the remaining staff of Inflection AI and its shareholders with a lucrative contract. 

Last week, Microsoft's acqui-hire of Stability AI's staff ran afoul of another major AI trend, the ever more intensive investigations by regulatory agencies in the U.S., the U.K., and the EU, who show increasing concern about anticompetitive GenAI  mergers and acquisitions by Big Tech. Indeed, two weeks ago Microsoft and Apple decided not to have seats on OpenAI's board of directors out of deference to the heightened concerns of the regulators

BackToTop

3) Coalition for Secure AI

A large segment of the AI community has long been concerned about p-doom, the small, but non-zero probability that GenAI might evolve into a phenomenon that inflicts catastrophic harm on humans. Recently, however, more and more people and more and more government regulators have been showing increasing concern about the much larger probability, nay, the near certainty that unethical users might use powerful proprietary GenAI tools developed by profit-oriented Big Tech corporations to inflict catastrophic harm on other individuals, on organizations, and on entire communities. 

Profit-oriented BigTech has resisted regulation with unflinching ferocity. Nevertheless, it feels compelled to provide assurances that its products are safe. So Big Tech gets together from time to time to reaffirm its commitment to "open-source" development. Last week Big Tech announced the formation of CoSAI, the Coalition for Secure AI. 

• "CoSAI is an open-source initiative designed to give all practitioners and developers the guidance and tools they need to create Secure-by Design AI systems. CoSAI will foster a collaborative ecosystem to share open-source methodologies, standardized frameworks, and tools."
  
  
• "CoSAI’s founding Premier Sponsors are Google, IBM, Intel, Microsoft, NVIDIA, and PayPal. Additional founding Sponsors include Amazon, Anthropic, Cisco, Chainguard, Cohere, GenLab, OpenAI, and Wiz".

Now we can all stop worrying ... for a few days ... :-)

BackToTo

B. Top 3 stories in past week ...  

1. OpenAI
   "OpenAI is releasing a cheaper, smarter model" -- GPT-4o mini, Kylie Robison, The Verge, 7/18/24 *** 
   -- This story also covered by Wired, TechCrunch (plus video), VentureBeat, Engadget, Wall Street Journal, .... and OpenAI
   
   
2. Public Policy
   "UK antitrust officials join FTC in investigating Microsoft's hiring of Inflection AI staff", Kris Holt, Engadget, 7/16/24 *** 
   -- This story also covered by TechCrunch, The Verge, Bloomberg, 
   
   
3. OpEds+Misc
   "The biggest names in AI have teamed up to promote AI security", Emma Roth, The Verge, 7/18/24 *** 
   -- This story also covered by  Google ... and CoSAI 

BackToTop