Interview/ Anne Neuberger, former deputy national security adviser for cyber and emerging technology, the US
HER GRANDPARENTS survived the Holocaust while her parents were among passengers saved by Israeli commandos in the Entebbe hijacking. Today, Anne Neuberger teaches the world how to survive the threats of the future, be it cyberspace or the questions raised by AI. A deputy NSA for cyber and emerging technology during the Joe Biden presidency, Neuberger is a distinguished visiting fellow at Stanford University. She spoke to THE WEEK about where India could find its sweet spot in AI. Excerpts:
Q/ We first thought AI was about information, about an easy way of doing things. Now we realise it’s going to change industry, economy, governments, everything. How do nations approach that?
I think countries are starting to grapple with the promise and the perils of AI. For example, when you think about the promise in the area of drug discovery, that’s huge. And when you think about some of the risks, it is likely that AI will accelerate and scale different kinds of cyber attacks and we have infrastructure that’s not quite ready for that.
So, I think countries are starting to think about how they promote the innovation, ensure that people get trained, get comfortable using AI, [and] ensure that regulations [are] updated.
Q/ AI is going to change the way wars are fought. That involves a major strategy reboot for governments. Do you think all nations are prepared for this?
First, AI changes intelligence. There’s so much intelligence collection and analysis, there’s so much data, and that’s traditionally been a challenge. You need to make sense of it. So, the goal is to [make] AI help countries generate intelligence faster and better to prevent attacks.
And then the second piece is, countries are starting to incorporate AI in their military operations. To coordinate a set of drones, for example, or to train them on where to go based on imagery, a location or a path.
Some of the tension is around what AI systems are allowed to do autonomously and how different countries make those decisions. In countries like India and the US, for example, as democracies, we say we want to make sure a human is in the loop.
Those are the areas where policies, international agreements don’t yet exist, [but] the technology is moving very quickly.
Q/ Cyber attacks are borderless. Now with AI, do we need a new kind of understanding between nations?
Digital space has no borders. And I think cyber [warfare] and digital space is going to change to being machine against machine. It is too fast and too scaled [to have] a human defender making real-time decisions against a system. So, as a result, I think what countries need to focus on is: [have] rules around not disrupting critical infrastructure in another country.
Q/ AI raises cybersecurity threats to a new level in a country like India with a massive internet user base, but which probably doesn’t have the safeguards of western Europe or the US.
The India stack (collection of government-operated digital infrastructure systems) is a real strength in AI, because you have a digital public infrastructure system where people can do secure transactions tied to identity. Those are foundational trust roots that you can build an AI stack on top of. So, that’s the first piece.
The second piece is educating the public to be both AI savvy for all the promise it has, but also to be aware, for example, about deepfakes. It will be very important that citizens trust our law enforcement systems.
Q/ How unequal is the world going to be with AI?
When Mukesh Ambani talks about competing on tokenisation so that the costs are [affordable] and with some of what India has done with the Bhashini language models (to translate Indian languages) with making health care and agriculture models available, the key will be what it takes to scale. I’m a big believer in small language models, because I think they’re more purposefully built.
And India built a communications infrastructure as a foundational piece to scaling that.
That’s the path India has decided to take. Particularly, for example, when you’re thinking about a farmer trying to diagnose crop disease, that’s rather focused and purposeful. When the approaches are people-based, we can measure progress by how much difference a particular AI application makes, and then [think of] how to scale it.