15.7 C
Tuesday, May 21, 2024

Another round of firings within Google’s artificial intelligence brain trust, as well as more discord

Following the dismissal of two researchers for criticising the biases built into artificial intelligence systems less than two years ago, Google has fired another researcher for questioning a paper the company had published on the abilities of an artificial intelligence specialised for chip-making called specialised artificial intelligence (SAI).

The researcher, Satrajit Chatterjee, led a group of scientists in calling into question a widely-publicized research paper published last year in the scientific journal Nature, which claimed that computers were capable of designing certain parts of a computer chip more quickly and effectively than humans.

A group of four people familiar with the situation who were not permitted to speak publicly on the matter said Dr. Chatterjee, 43, was fired in March, shortly after Google informed his team that it would not publish a paper that rebutted some of the claims made in the Nature journal, according to the sources. According to a written announcement from Google, Dr. Chatterjee’s employment has been “terminated with reason.”

Google refused to provide any details on Dr. Chatterjee’s departure, but it issued a vehement defence of the study he critiqued as well as its refusal to publish his judgement.

In spite of the fact that Google has highlighted the technology’s promise, it has experienced pushback from workers about its implementation. Concerned that the company’s artificial intelligence (A.I.) may end up murdering humans, Google workers demonstrated against a contract with the Department of Defense in 2018. Google finally decided to withdraw from the initiative.

Timnit Gebru, a leader of Google’s Ethical A.I. team, was fired in December 2020 after she publicly criticised the company’s approach to minority hiring and pushed for the publication of a research paper that pointed out flaws in a new type of artificial intelligence system for learning languages, among other things.

Dr. Gebru was attempting to get permission to publish a research paper about how artificial intelligence-based language systems, such as Google’s technology, may end up using biassed and hateful language that they learn from reading text in books and on websites before she was fired. Dr. Gebru was fired shortly after requesting permission to publish her paper. In an interview with Dr. Gebru, she expressed her frustration with Google’s reaction to such objections, which included the company’s unwillingness to publish the study.

The Nature research, which was published in June, advocated for a method known as reinforcement learning, which the authors said might be used to enhance the design of computer processors. A breakthrough in artificial intelligence was heralded, as was a significant advance over previous techniques to chip design, as a result of the technology. A Google spokesperson said that the company utilised this technology to design its own chips for artificial intelligence computer applications.

Google had been working on bringing the machine learning approach to chip design for years before publishing a report that was almost identical to this one a year before. A few months later, Google reached out to Dr. Chatterjee, who holds a doctorate in computer science from the University of California, Berkeley and had previously worked as a research scientist at Intel, and asked him to investigate whether the approach could be sold or licenced to a chip design company.

Although Dr. Chatterjee raised misgivings in an internal email about some of the paper’s claims, three persons who spoke with him said he also questioned whether the technology had been adequately evaluated before publishing.

One of the research’s two principal authors, Anna Goldie, who developed the paper with a colleague computer scientist, Azalia Mirhoseini, claimed the alterations from an earlier version did not need the publication to go through the complete approval procedure. Dr. Chatterjee and a small group of internal and external researchers were given permission to work on a study that called into question several of Google’s assertions.

In their rebuttal article, the researchers expressed their desire to take the matter to Mr. Pichai and the Alphabet board of directors for further discussion and consideration. They said that Google’s choice not to publish the reply was a violation of the company’s own artificial intelligence principles, which include sustaining high scientific standards of excellence. According to the persons who spoke with me, Dr. Chatterjee was notified shortly after that he was no longer employed.

She said that Dr. Chatterjee had approached them about managing their project in 2019, and that they had rebuffed his offer. She said that when he subsequently questioned it, he was unable to provide proof to support his claims and disregarded the information that had been offered in response.

She said that the paper has been peer-reviewed by Nature, which she described as “one of the most prominent scientific journals.” She went on to say that Google had utilised their technologies to develop new chips, which were presently being employed in the company’s computer data centres.

Following the distribution of the rebuttal article to academics and other professionals outside of Google, the dispute expanded across the worldwide community of researchers who specialise in chip design and fabrication.

Some analysts are sceptical what Google’s study would imply for the broader tech sector, despite the fact that the chip producer Nvidia claims to have utilised processes for chip creation that are comparable to those used by Google.

David Faber
David Faber
I am a Business Journalist of The National Era
Latest news
Related news


Please enter your comment!
Please enter your name here