India is well-positioned to unlock gen AI's potential: Google AI chief Jeff Dean

Though Gen AI has great potential to unlock so many good things, it is not perfect. However, "there is progress in factual accuracy compared to these models a year or two ago", Dr. Dean noted.

[From left] Manish Gupta (Director of Google Research India), Grace Chung (Google Research Australia Chief) and Dr. Jeff Dean (Chief Scientist at Google DeepMind and Google Research) at Google @Research event in Bengaluru on February 1, 2024.

Credit: DH Photo/KVN Rohit

Since the advent of personal computers in 1970s, interacting with machines has changed steadily, and has been more drastic in the last few years. Earlier, people had to type commands to perform tasks and dial in numbers to make a call. Now, in this modern era of Generative Artificial Intelligence (Gen AI), users can just speak naturally to the device to hail a cab or churn out beautiful photos from thin air. People can even ask gen AI bot to instantly create an application.

While there are fears of Gen AI taking away jobs, it also holds great potential to change people's lives for good in healthcare, education, agriculture and several other fields. This and many others were the talking points at Google's first-ever 'Research@' event in Bengaluru on Thursday (February 1).

Dr Jeff Dean, chief scientist at Google Research and DeepMind graced the event. He is very optimistic about India, and how it can play a crucial role in unlocking Gen AI potential for the world.

“India has tremendous technical talent; there's a long tradition of computer science students in India doing amazing things. Anyone with an interest in computer science is turning their attention to machine learning and what it can do. India is well-positioned in this area because many students are already in this field and people are already in this industry. It's an important thing across the world,” Google AI chief Dr. Jeff Dean said.

Dr. Dean and Oriol Vinyals are co-leading Google DeepMind's ambitious Large Language Models, which power the newly launched Gemini AI. The new gen AI can under complex queries in text, code, audio, image, and video.

Depending on use cases, Google is offering Gemini in three forms-- Nano, Pro, and Ultra with different capabilities.

Gemini Nano and Pro models are already deployed on smartphones (Pixel and Samsung Galaxy S24 series) and early response among customers and critics has been great.

[Note: More Android phones will get support for Gemini AI models later this year]

In India, Google Cloud is also collaborating with Bhashini within Ministry of Electronics and Information Technology(MeitY) to launch a Center of Excellence on Generative AI and Language Inclusivity. It aims to equip one million professionals and students with generative AI knowledge and skills so they can thrive in this new AI era.

“One of the efforts I helped encourage was our 1,000 language effort, aiming to make AI models available in all major languages. This is crucial because many languages spoken by tens of millions of people lack extensive written material. This is important because making these languages accessible to AI technologies will unlock its potential for different communities that have been previously underserved,” Dr. Dean noted.

Though Gen AI has great potential to unlock so many good things, it is not perfect. Like humans, it is susceptible to fallacy too. With time and new emerging data sets, the large language models can be refined to make them objective and neutral.

"It is really important to be aware of the rough edges and potential pitfalls that arise on the scaling of machine learning models. There is a lot of progress in factual accuracy compared to these models a year or two ago. It is important, we continue to work on maximizing the benefits while actively mitigating the shortcomings,” Dr Jean noted.

Get the latest news on new launches, gadget reviews, apps, cybersecurity, and more on personal technology only on DH Tech.