Interview: Australian Financial Review

Dr Alan Finkel was interviewed by Ben Potter for the Australian Financial Review about artificial intelligence, energy and his role as Chief Scientist.

You can read the full article below, published in the Australian Financial Review on 17 November 2017.

I have been chatting with Alan Finkel about artificial intelligence for 18 minutes when the Chief Scientist invites me to make a “substantial commitment” to science. I just have to have my head lopped off and my warm brain removed so that it can be cut into invisibly thin slices to be analysed by electron microscopes and fluorescent imaging devices many times more accurate than an MRI.

“And from all this information we would reconstruct a three-dimensional chemical, electrical and morphological picture of your brain, and then if we know how we could interpret that as being you,” Finkel says. “And then if that worked you would then exist into the future, the indefinite future, in digital form, but you would have no physical manifestation left because we sacrificed you … And if it didn’t work it would be a sad day in your life.”

We are sitting at a window table at Becco, an Italian bistro in Melbourne’s Crossley Lane, and I think I see what he means. For entree Finkel has chosen eggplant parmigiana, and I am chomping on grilled asparagus with crumbed egg. We are drinking water. Finkel is unfailingly proper, and precise but detailed in conversation. He’s an incurable optimist with steel behind the polite persona.

Science fiction enthusiasts will instantly know that he is alluding to “the singularity” – a future time when artificial intelligence (AI) and machine learning surpass human capabilities in an explosive cycle of accelerating progress, conjuring up fears of machines taking over and relegating humans to the status of pets.

Finkel isn’t alarmed – he calls singularity movies and novels dystopian flights of fantasy. Rather, like Elon Musk and Stephen Hawking, he says we need “adaptive” and “anticipatory” regulation to cope with AI’s breathtakingly rapid development as software mimics the brain and learns from experience.

A few years ago, he wrote in The Incurable Engineer – his column in Cosmos, the science magazine he and his wife Elizabeth founded in 2005 after selling Axon Instruments (the Silicon Valley company Finkel had started in the early 1980s) – that AI had a long way to go to achieve reliable facial recognition or voice recognition. A Lexus he had about six years ago had a voice-activated phone, but got it wrong as often as not – a failing of the technology, not the carmaker.

Now he instructs Siri to send himself an email or the voice-activated GPS on his Tesla to take him somewhere, and the performance is vastly improved.

“I love it – I get in my car and I press the button, and I say, ‘Car, take take me to the underground car park at Parliament House’, the screen transforms itself and I’m on my way’.”

But just as we’ve learnt to live with  beneficial but dangerous technologies such as motor cars and electricity, Finkel says, we need a set of rules for AI that can work with industry and evolve with the technology.

“If we do that and we do it well, we’ll have a growing industry in AI that continues to make our lives more comfortable, that enables us to amplify whatever it is that we do in our home environments, in our work places. And if we don’t, we’ll have AI that is perceived by many to be dreadful, and then you’ll run the potential of a backlash.”

Germany’s Bundestag has passed a rule requiring autonomous vehicles to treat all human lives equally and minimise the loss of life when collisions are unavoidable. Agree or not, that’s a rule that will allow autonomous vehicles to advance, Finkel says.

Finkel has spent a lot more time in the past year pondering Australia’s energy market than AI. After South Australia’s blackout on September 28 last year, federal Energy Minister Josh Frydenberg asked him to lead a review of energy security. They gave their 50 recommendations to the government in June, to the acclaim of industry and states, and almost all were accepted.

But the heart of the report – the “orderly transition” to a lower-carbon energy system – was rejected by Turnbull government conservatives because it was too “anti-coal”. It included an agreed long-term CO2 emissions trajectory, a “Clean Energy Target” and a reliability obligation for new wind and solar farms. Coal fans queried his credentials and “magic pudding” economics. Now energy policy is hostage to the parliamentary citizenship debacle.

How does it feel? Finkel says he is a glutton for punishment and enjoys working hard with capable people to achieve important things. Pressed, he says he is personally “saddened that a good solution has not been rapidly adopted because if it had been we already would have been better placed with respect to pricing and security for electricity”.

But he is not frustrated or angry or distressed because it was a “true privilege” to have led the review and turning them into law is the job of government.

Pull the other one? “I am frustrated,” he owns – not by the Clean Energy Target’s dismissal, but that the review’s narrative “has not been clearly understood by a large majority of the press, government and broader population”.

The review was thorough – it consulted energy market players in many European and US energy jurisdictions.

“There’s no doubt that they are all fully cognisant of the fact that the world of energy is being subject to rampant disruptive technological change,” Finkel says.

“There is no way to stop that and what’s essential is to get to the future harmoniously with that technological change, which means at the best possible cost, with strong stability and taking advantage of that technological change to lower your carbon emissions.”

The review’s report was a model of clarity and concision for such a big topic, but former prime minister Tony Abbott boasted of not having read it.

Lower emissions are the end goal, the energy mix is just an input, Finkel says. “I think we have delivered a fully optimised package through the 50 recommendations of the Finkel review.

“The recommendation they didn’t adopt is recommendation 3.2 for the orderly transition and that is the strategy for the energy sector so it’s very significant that it was not adopted,” he says. “It had support from the industry but not from the government.”

He notes that the Energy Security Board had adopted the same three-legged framework for the alternative, politically palatable National Energy Guarantee (NEG) that Frydenberg asked it to devise – but with different details.

“It’s significantly less ambitious than the modest trajectory that we modelled,” Finkel says. And more complicated than the Clean Energy Target, which was modelled on the well known RET.

“I would say it’s difficult to imagine anything simpler than the CET as a mechanism,” he says.

But he adds that the NEG is workable, and with intensive consultation with the industry can deliver the strategy that we need, and meet any future changes in the emissions target.

Finkel the engineer doesn’t understand why anyone would fixate on the shares of wind, solar or coal power in the energy mix rather than the end goal of cutting CO2.

“As it happens the easiest, lowest cost way to do that is with wind and solar.”

Energy has obscured Finkel’s day job – Chief Scientist – but that may change with the release of Innovation and Science Australia’s 2030 strategic plan. It’s nearly done: “watch this space”, says Finkel. Turnbull’s enthusiasm for innovation had to be tempered during last year’s election because regional Australians saw it as a threat to their jobs.

Finkel says the situation is not as dire as painted by some data, and points to the 26 years of continuous economic growth, the technological sophistication of our world-beating iron ore industry, tech successes like Atlassian and growing industry collaboration and startup “accelerators” in universities and CSIRO.

But the ASX’s failure to regain pre-global financial crisis highs – the digital tech-powered S&P 500 and the Dow have hit new heights – illustrates the shortfall in what Finkel calls “Silicon Valley tech” businesses, despite pockets of notable success.

“It’s not a zero, but it’s not a sector of our economy where you would say we’re kicking goals – and we should be, because it’s a growing sector, and many people say IP [intellectual property] is the new iron ore.”

Finkel stumbled on making scientific instruments when he was studying for his doctorate in electrical engineering at Monash University, and the head of the department wouldn’t let PhD students buy electrical kit. He had to build his own instruments for measuring tiny electrical currents in the brain.

At the ANU, he did post-doctoral research in neuroscience and found he was better at making instruments than research. He met Elizabeth, followed her to San Francisco where she was doing her own post-doctoral work in molecular genetics and launched Axon. He told investors and managers that if an engineer announced he or she had delivered a project on time and on budget “I’d fire them – because they weren’t ambitious enough”.

After selling out of Axon, Finkel became an advocate for science and education, not just by launching Cosmos – “the best science magazine in the universe” – but also by becoming Chancellor of Monash University and chief technology officer for Better Place Australia, the short-lived electric vehicle charging company that was too far ahead of its time.

Along the way he acquired a taste for climbing serious mountains – tackling Kilimanjaro in 2003, Mont Blanc with his two sons eight years later, and Mera, a 6476 metre Himalayan peak, in 2014. But he’s now done with that.

His next adventure is a ticket for him and his two sons and Cosmos’ founding editor, Wilson da Silva, to space on Richard Branson’s Virgin Galactic – which has an indeterminate launch date.

Why? “Why wouldn’t you want to do it? How exciting to be up there literally looking down on earth and seeing the blue fringe of the atmosphere from above … and floating.”

When a headhunter called him in 2015 about the job of chief scientist, he had four names to recommend to her, and his own plans. But she said his was the name on everyone’s lips, so he consulted Elizabeth. “She said, ‘Alan you should do it – your country needs you’ … And it’s not often you get that from your wife, so I thought, ‘Well, that’s a sign’.”

When I lazily venture that theirs must be a great partnership after “50-odd years”, he instantly counters: “No, no, no – it’s not that, it’s 35. If you said that, I’d be in such trouble.”