This layer comes out today’s AI landscape – what do you know – can be traced to fear, jealousy and strong capitalist ambitions. Emails detected Justice Department antitrust case against Googlefirst informed by Business Insider, show Microsoft executives expressing excitement and envy over Google’s leadership in artificial intelligence. This spurred the urgency that drove the Windows maker billion dollar initial investment now in OpenAI, an indispensable partner.
In a heavily redacted 2019 email titled “Thoughts on OpenAI,” Microsoft CEO Satya Nadella sends a lengthy message from CTO Kevin Scott to CFO Amy Hu. “A very good email explaining why I want us to do this … and why we’re going to have our infra people execute,” Nadella said.
Scott wrote that he is “very, very concerned” about Google’s rapidly growing AI capabilities. He says he initially rejected the company’s “game-playing stunts,” referring to Google’s AlphaGo models. One of them Go defeated the world champion Ke Jie In 2017, it was a remarkable achievement at the time. (Google’s later models surpassed it, reduces the need for human training in general.)
But Scott says it was “a mistake” to remove Google’s gaming progress. “When you take away all the infrastructure they built to build [natural language] models we couldn’t easily replicate, I started taking things more seriously,” Scott wrote. “And I’ve been very, very concerned about trying to figure out where all the capability gaps are between Google and us for model training.”
Scott explains how Microsoft has struggled to copy Google’s BERT-lage, an artificial intelligence model that deciphers the meaning and context of words in a sentence. Scott blamed the infrastructure leaps his rival made — and Microsoft didn’t.
“It turned out that it was not easy for us to repeat the BERT-lage. Although we had a template for the model, it took us ~6 months to train the model because our infrastructure was not up to the task,” the Microsoft CTO wrote. “Before that, Google had BERT for at least six months, so in the time it took us to crack the ability to make a 340M parameter model, we had a year to figure out how to get it into production and move on. larger scale, more interesting models.
He also admired and envied Google’s Gmail autocomplete capabilities, saying it was “getting a lot better.” He commented that Microsoft was “several years behind the competition in terms of [machine learning] scale.” He commented on the “exciting” growth of OpenAI, DeepMind and Google Brain.
Scott noted the “very smart” people in Microsoft’s machine learning teams, but said their ambitions have been curbed. “But the core deep learning teams at each of these big teams are very small and their ambitions are limited, which means that even if we start feeding them resources, they still have to learn to scale. Scott wrote. “And we’re several years behind the competition in terms of ML scale.”
The company complied with the CEO’s wishes after Hudu said Scott’s concern was “why do I want us to do this,” i.e. invest in OpenAI. Microsoft invested a billion dollars in the Sam Altman-led startup in 2019, and the rest is fast-changing history. (It is now invested 13 billion dollarsn.) This is technology does some incredible things but threatens closes the labor market and preachers give their most powerful tools to date it was already a year widespread misinformation.
