Microsoft's Sandeep Alur on the concerns raised by Generative AI

Some days ago, we were seeing the image of Joe Biden and Donald Trump meeting in the White House while in reality, Trump was in a different city. While this is not the first instance of generative AI creating a stir and raising concerns, it also outlined the potential of generative AI. Here, Sandeep Alur discusses the several concerns AI has been raising.

Sandeep Alur, Director - Microsoft Technology Sector, Microsoft, talks about the three elephants in the room raised by AI - Plagiarism, Language Availability, and AI being adversary and not assistive.

Plaigarism By Generative AI

Plagiarism has been one of the biggest concerns revolving the generative AI technology, whether it's accidental, incidental, or intentional. Alur mentions this technology is based on knowledge that is already publicly available. This means that the technology at this point has a tendency or a possibility to repeat what somebody has already created with a probability of that creation garnering more popularity than the 'original' creation. This would also happen at a faster rate, as "machines can do everything faster."

Language Barriers

In January 2018, Microsoft initiated the potential of Artificial Intelligence and Deep Neural Networks to improve real-time language translation for Hindi, Bengali, and Tamil. Deep Neural Networks-powered language translation, generates more accurate and natural-sounding results. The benefits of AI-powered are not just restricted to translation or the Microsoft Translator app. It could be helpful for internet browsing on Microsoft Edge browser, on Bing search, Bing Translator website, as well as Microsoft Office 365 products like Word, Excel, PowerPoint, Outlook, and Skype, along with APIs on Azure that they can use in their products.

Also Read: What lies ahead for advertising as AI and influencer marketing gain more prominence?

Discussing the language barriers for AI-powered technologies that impact the progress towards having AI-powered technologies across (for example) the 22 major Indian languages is functional accuracy, he describes micro-service in the world of innovation being an underlying model which is dependent on an open source having a median-userbase of English-language speakers. The accuracy of these models in surpassing human parity is dependent on investment into creating a one-stop-shop, that is backed by evidence of its accurate performance. For now, we have that evidence for English.

Assistive Technology

Generative AI technology has harmed employment in several industries and replaced various entry-level jobs. This raises the concern of AI being 'assistive technology or adversary technology.' Alur responds to this concern with examples such as Copilot, a cloud-based AI tool developed by GitHub and OpenAI to assist users of Visual Studio Code, Visual Studio, Neovim, and JetBrains integrated development environments by autocompleting code.

The use of AI technology is collaborative - a human persona dealing with a digital persona. India has a massive database, but lacks skilled talent for data management and development, across the board. Project flow is hampered due to these reasons and often professionals are not able to meet deadlines. But with AI one can do things faster, and take on projects that are empowering developers to stay productive to meaningfully. It's like having an assistant that can help unlock our potential in multiple avenues.

Sandeep Alur, Microsoft spoke at FICCI Frames 2023. This article contains insights shared first at the event.