We're excited to introduce Dr Gopichand Katragadda as the IET's 142nd President!
A seasoned technologist and executive leader, Gopi is the Founder of Myelin Foundry – an AI company with a vision to transform human experiences and industry outcomes. He’s also an Independent Director of Bosch India Limited and ICICI Securities and a member of the NASSCOM governing council for the Centre of Excellence for Data Science and AI.
Gopi’s presidential address, Engineers and AI: the key to delivering a resilient future, will take place at 6.30pm GMT on Tuesday, 21 November 2023.
You can join us in-person at IET London: Savoy Place, or tune in online via YouTube. Register online and find out more via our website.
Ahead of the talk, we caught up with IET Head of Technology Strategy and Product Solutions, Dr David Smith, to get his views on this exciting and controversial subject area...
Thanks for your time, David. Let’s get stuck straight in – what's your favourite AI innovation to date?
I probably should timestamp these answers so folks can know when, in terms of AI developments, I was writing (September 2023). I’m an ex-molecular biologist (a long time ago, in a galaxy far, far away) so, for me, the work where AI has made incredible strides into understanding the structures of proteins from their amino acid sequences has me going ‘whoa!’ because that will lead directly to vital drug development and other health benefits for humanity. We also now have version one machines that you can actually converse with for extended periods of time (not just quick questions), and in a manner remarkably like how we talk to each other. That’s incredible.
How ‘intelligent’ is AI in its current state?
A very good question indeed! So we are nowhere near so-called artificial general intelligence (useful definition here: Artificial general intelligence - Wikipedia). It’s important to know that no matter how ‘human’ these AI’s sound (or read?) they do NOT have agency. Having said that… GPT4 in particular, seems to have some quite impressive Theory of Mind capabilities and other logical reasoning skills. And we should expect these to continue to improve – at speed. We should also remember that inside of these AIs is some sort of representation of most of human knowledge and culture from the last 16 or so years. In English, of course, with a bias towards western culture. And these version one AIs are showing that just like the first PCs were general purpose machines – we have AIs with a remarkably broad set of skills and capabilities, even if some of them are bit rough around the edges.
This year’s President’s Address is themed around AI, as mentioned above…what part of the talk are you most excited about?
I’m excited for a few reasons. I’ve had the privilege of travelling to India for some 12 years now. I’ve seen Mumbai change in that time, I’ve seen Bengaluru change in that time. So to have our President come from the worlds largest democracy is an important moment for us as an institution. We will get a different perspective on world matters; on engineering matters. I saw some of that last year at the inspiring Future Tech Congress in Bengaluru. I’m looking forward to hearing Gopi’s perspective on AI.
Are there any industries you feel AI will affect the most?
A better question may be what won’t be affected! I suspect that physical skills and capabilities are the ones that are probably safest, as right now we don’t have good foundational AI for the control of robots – because the data doesn’t exist to enable the current advances in AI to be applied to that space. Right now, anyway… So, while there’s no robot surgeons in hospitals at the moment, I wouldn’t bet against machines that help surgeons do ultra complex surgery in the future – as an example. Healthcare
The rise of the machines… Putting specific industries aside, do you believe there is space for AI in the broader workplace? Can a balance be achieved?
This comes down to a question that every organisation will have to answer for themselves – what are our principles? Very broadly – the IET is in the information business. We help people in engineering from kids in school all the way up to advisors to governments, with everything in between. I feel our use of AI should therefore be about moving our organic-based large language models (see what I did there ) up the value chain of what we do. So AI should be a force multiplier for us, not a replacement for us.
And what about soft skills... could AI replace those?
Could one conceive of an AI that could help people struggling with cognitive decline? Helping them remember or replay treasured moments? Can AI ever get angry? Tired? Always there, happy to sing a song again and again? Helping to communicate with other humans? Amazon’s Alexa turned out to be very helpful there, and that was very crude by today’s new standard, so there’s certainly a place for AI in the world of soft skills. But it needs to be the right place.
So, do you think AI can realistically ‘replace’ humans?
We have robots driving and indeed flying over the surface of Mars! We have a ‘grocery store’ who builds big warehouses populated by robots that can move in three dimensions to fulfil food orders (and yes that company is SUPER KEEN on autonomous delivery vehicles), and neither of those two examples yet make use of newer developments in capability. The history of technology adoption has followed the ‘J-Shaped curve’ in that they don’t immediately translate into productivity growth (so more opportunities for humans). Very simply – that’s because everybody is first reinventing stuff to take advantage of the new tools, and then boom! The curve races off up and to the right at an increasing rate. Sometimes they stall out (remember the Dotcom boom of the early 2000s) and then another wave comes in. But what if this time it is in fact different? Something to think about…
What’s the IET doing to develop our understanding of AI?
We’re doing a couple of things right now. Firstly, we’re making sure that we’re spending the time to get to grips with what’s happening here and how we can build the frameworks to be able to take advantage of it. As the saying goes, time in reconnaissance is time seldom wasted. The second thing is to start a transparent conversation so that all parts of the IET can grapple with what it means for them. We held a recent AI workshop for colleagues to explore opportunities across our many products, services, and areas of expertise – so that was a great first step.
Do we need to better control the rate that AI is advancing? Is it getting out of hand?
There’s a rich history of technology doom-mongering. The speed of cars will kill us! Books - in the hands of the people?! – madness! etc… In my time in tech, I’ve never seen a rate of change quite like this. But I think the regulation issue isn’t about the tech per se, it’s actually about the organisations who are building them and the interactions between these organisations and the governments that should manage these things for the benefit of human society. There’s a wider point here; Chemistry had its reckoning in the horrors of World War 1; Physics with Oppenheimer and the A Bombs, Little Boy and Fat Man; Biology has had to reckon with the horrors of Eugenics… In each of those cases – regulation and governance followed in order to pull back from some very dark futures. Has computing had its Oppenheimer moment? My view is that ‘move fast and break things’ is not the approach to be tolerated here.
We have a 2030 Strategy in place to guide our direction as an organisation. Where do you think AI will be by then?
Humans tend to overestimate short term progress and badly underestimate long term progress. Six years out from now, where will things be? Well, GPT4 will be looked back on (hopefully fondly) as the Nokia 8110 of AI (look it up folks, then go watch the movie again). To paint a positive future, let’s imagine we’ve sorted out the data issues and therefore we’ll will have a digital butler/assistant of some sort who we can trust to help use with our daily lives. It’ll be partly with us on a mobile device, and partly cloud-based – and we will have agency over it. And if we let it, it’ll probably remember (wholesale) all our interactions in order to better understand us and what we are wanting from it. It’ll plug into services that are offered (perhaps specialist services that say a Professional Engineering Institution can offer – highly trusted and all that) to help it do these things. It will therefore be incredibly powerful… And with great power comes great responsibility.
On that final thought, then, what do you think is the future of AI specifically in engineering?
I’m very excited to find out! I think it will be revolutionary in helping to solve all sorts of very tricky engineering problems (say, how best to optimise power usage across an entire country, or how to manufacture new electronic capabilities (fun fact – AI is already being used to advance chip designs… for better AI capabilities!)). If we get the regulation right, the future is very bright indeed. But let’s not forget a key challenge here – the climate emergency. Our future rests on our ability to undergo the greatest technology change since The Industrial Revolution, in a fraction of the time, without destroying our planet in the process.