(This article is part of Today’s Cache, The Hindu’s newsletter on emerging themes at the intersection of technology, innovation and policy. To get it in your inbox, subscribe here.)
Google’s next AI project
Google is getting ready to display a product code-named Project Jarvis and possibly the next iteration of the Gemini model, reported The Information, citing anonymous sources. Project Jarvis could potentially take over a user’s web browser in order to help them complete tasks that would otherwise require multiple steps and separate processes. While AI models so far have been developed and integrated into the web experience to provide summaries and answer questions, Project Jarvis could take this a step further. Others working on similar concepts include Anthropic, which is backed by Google, and OpenAI, which is backed by Microsoft. Google is yet to confirm the product or its release timeline.
However, there are a number of issues that remain to be solved, including the question of how AI-powered web tools will treat paywalls, as well as the question of to what degree they will be monetised with advertisements and sponsored entries.
OpenAI’s Whisper hit with complaints
Though OpenAI has strongly endorsed its AI-powered transcription tool that is commonly used in hospitals as well as call centres, users are complaining that Whisper invents text that no one has actually spoken, leading to potential repercussions. OpenAI has warned against using its tools in high-risk situations but Whisper is still used in medical set-ups to help doctors record and review their patients’ information. However, computer scientists noted 187 hallucinations in over 13,000 clear audio clips that were studied.
Hallucination is an ongoing challenge that AI scientists are looking to solve, since models that hallucinate greatly increase the risk of running serious errors and biases when deployed in real life. However, experts are especially concerned that the ramifications of using Whisper in hospitals might not even be fully known because of medical confidentiality.
Microsoft fires two employees
Microsoft has fired two employees for holding a vigil at its headquarters in order to mourn the Palestinians killed by Israel in Gaza. The lunchtime event took place in Microsoft’s Redmond campus in Washington. However, Abdo Mohamed and Hossam Nasr were later fired. Both were also part of the “No Azure for Apartheid” coalition of employees who are against Microsoft’s sales of tech to the Israeli government as it carries out military action in Gaza, killing and displacing thousands. Microsoft confirmed that some terminations had taken place but cited its internal policies and privacy as it declined to offer more background about the action.
Pro-Palestinian tech workers in the U.S. have in the past come together to draw attention to how their companies supply high-level tech to Israel despite its ongoing war. Google had also fired employees over sit-ins against its relationship with the Israeli government.
Published – October 28, 2024 11:40 am IST
