In an email that was sent, CEO Satya Nadella expressed the company’s “extreme worry” on the capabilities of its competitor.
It has been discovered that the current state of the artificial intelligence landscape can be linked back to fear, envy, and tremendous economic ambition. What could you possibly know? Business Insider was the first to publish on emails that were discovered during the Department of Justice’s antitrust case against Google. These emails show officials from Microsoft expressing concern and envy over Google’s leadership in artificial intelligence. This prompted a sense of urgency, which ultimately resulted in the Windows manufacturer making its first billion-dollar investment in OpenAI, which is now a vital partner.
Microsoft CEO Satya Nadella forwarded a lengthy message from Chief Technology Officer Kevin Scott to Chief Financial Officer Amy Hood in an email thread that was highly redacted in 2019 and headlined “Thoughts on OpenAI.” The email that Nadella sent was a “very good email that explains, why I want us to do this… and also why we will then ensure that our infrastructure folks execute,” he added.
It was written by Scott that he was “very, very worried” about the rapidly expanding artificial intelligence capabilities of Google. His statement indicates that he first disregarded the “game-playing stunts” of the corporation, most likely alluding to the AlphaGo models developed by Google. In 2017, one of them achieved a spectacular victory by defeating Ke Jie, who was the world champion of Go at the moment. (Subsequent models developed by Google eventually beat that one, eliminating the requirement for human training entirely.)
However, according to Scott, dismissing the progress that Google has made in game-playing “was a mistake.” “When they took all of the infrastructure that they had built to build [natural language]models that we couldn’t easily replicate, I started to take things more seriously,” Scott wrote. “Those models were difficult for us to replicate.” “As I proceeded to investigate the matter in order to ascertain the locations of all of the capability gaps that existed between Google and us in terms of model training, I became extremely concerned.”
An artificial intelligence model that analyzes the meaning and context of words in a sentence is called BERT-large. Scott describes how difficult it was for Microsoft to replicate Google’s BERT-large. Scott placed the responsibility on the infrastructure advances that its competitor had made, whereas Microsoft had not made any such leaps.
It turns out that simply reproducing BERT-large was not an easy task for us to accomplish. In spite of the fact that we had the template for the model, it took us around six months to train the model because our infrastructure was not capable of handling the task,” the Chief Technology Officer of Microsoft noted. According to the statement, “Google had BERT for at least six months prior to that,” which means that during the time that it took us to cobble together the capabilities to train a 340M parameter model, they had a year to figure out how to bring it into production and to go on to larger scale models that were more intriguing.
In addition, he loved and envied the auto-complete features of Google’s Gmail messaging service, stating that it was “getting scarily good.” He made the observation that Microsoft was “multiple years behind the competition in terms of [machine learning]scale.” He remarked on the “interesting” development of OpenAI, DeepMind, and Google Brain after observing their rise.
Despite the fact that Scott praised the “very smart” individuals who work on Microsoft’s machine-learning teams, he stated that their objectives were limited. According to what Scott mentioned, “But the core deep learning teams within each of these larger teams are very small, and their ambitions have also been constrained.” This suggests that even if we are beginning to feed them resources, they still need to go through a learning process in order to scale up. In addition, we are several years behind the competitors in terms of the size of machine learning.
Following Hood’s insistence that Scott’s concerns were “why I want us to do this,” which is another way of saying invest in OpenAI, the company complied with the intentions of its CEO. In 2019, Microsoft made a one billion dollar investment in the firm that was run by Sam Altman, and the rest of the history is a history that is always evolving. It has now invested thirteen billion dollars. It is a technology that is capable of doing some great things, but it also poses a threat to the labor market and provides propagandists with the most powerful weapons they have ever had at their disposal in an age that was already characterized by widespread deception.