The NBCS AI Summit was a valuable day for students, academics and industry professionals

Congratulations to Hannes Ennemann, Jedd de Gouveia and the team of students, who, supported by NBCS staff, conceived and carried off this wonderful, fruitful summit. We were very pleased to be able to support students in this way, and grateful to have the calibre of academics and industry professionals who contributed across the day, including co-host, Professor Danny Liu of the University of Sydney. I learned lots.

The following was my address to the summit:

Good morning, everyone and welcome, distinguished guests and distinguished students, welcome one and all. I hope that today is a productive and mind-bending day for all of us.

I want to start by recognising how little I know. I am reasonably smart, reasonably attentive, I read a lot and live in the middle of the information revolution. I am part of a world where we have more access to more information more easily than any people who have ever lived. It’s exhilarating and exhausting at the same time.

I’m a history teacher, so I am acquainted with things that have changed the world before now. Farming, the wheel, aqueducts, bronze and iron, mechanisation, water and steam, industrialisation, specialisation, factories and trains, machinery, the bicycle, the car, antibiotics, the calculator, computer, the internet. All good things, but none of them always good or only good.

And now we sit at the dawn or in the morning of a new age. But permit me to express the scepticism of someone who is a bit older, hopefully a bit wiser, or at the very least someone who has a longer-term perspective. We often overestimate the initial impact of change and underestimate the long-term impact. I think that this will probably be true for AI.

Here’s my word of caution. If you think that LLMs will lead to AGI, I think you are wrong. Maybe AGI will be achieved through neuro-symbolic AI because neuro-symbolic approaches combine logic, knowledge representation and reasoning as well as neural networks. Without the rules, parameters and guidelines of the symbolic approach, the neural networks alone are getting further away from the goal of AGI.

Given that LLMs are massive predictive text machines, the quality rather than the quantity of their input will determine the quality of their output. Given that they continue to be fed the internet and its content, it should not surprise us that they just can’t win.

Trillions of dollars are being burned because too many people have been convinced that brute force will “crack the code”. It won’t. But what will? Logic, carefully crafted rules and representations. Very specific use cases and training within them. I think that AI will, over time, become more effective as it becomes more field specific, whether branches of medicine or law or theology, maths or business.

The internet has led to the great democratisation of information. This seems like a good thing. And it has been. But disinformation and misinformation are now more prevalent than ever. Will AI follow a similar pattern?

What are we trying to achieve? What is the desired outcome? Will it be like so many of the machines of the past? Will it be like the washing machine and the electric oven – labour saving devices which save time that we now spend on other things. Will the two-hour workday become a thing? I doubt it.

Will AI be a force for good – expanding the capacities and opportunities for us all or is it another way for billionaires to do away with their knowledge workers in the same way that automation reduced the need for manual workers? What skills will be irreplaceable? As Nassim Taleb writes, “if you don’t want to be replaced by a computer, don’t write like a robot.”

Will AI be a force that levels the field for everyone or tilts it further away from the ordinary person? I am excited at what AI might bring, but cautious about unintended or unforeseen consequences. Like much that has gone before, AI will prove to be a wonderful servant and a lousy master.

Don’t outsource your thinking and don’t outsource your learning. Without the ability to think and to learn, you will become captive to forces beyond yourself and diminished as a human. As Andreas Schleicher writes ‘Learning does not happen when the machine does the thinking for you.’

To use AI well, we should follow the advice of Benjamin Laker, “Automate tasks, not trust, use AI to widen rather than narrow your perspective, and use it as a sparring partner rather than a cheer leader.”

Noted AI expert and sceptic, Gary Marcus writes, “GenAI has been undermining education, opening up mass surveillance, increasing disinformation, delusions, impersonation, phishing, and other forms of cybercrime, nonconsensual deep fake porn, bias in employment, economic disparity, drowning the world in slop and unwanted, over-leveraged environment-damaging data centres that risk causing a recession. Simultaneously it has empowered a bunch of people who want to privatize almost all the gains while leave all the downsides to society, taking almost zero responsibility. I don’t think we are better off than we were four years ago.”

The question for us today is how we avoid this and how we put a use case for AI that will leverage its strengths for good and for all.

I remember when America invaded Afghanistan in the early 2000s and American pilots were flying unmanned planes dropping bombs on Afghan villages. The planes took off from US bases in Afghanistan, but the pilots were in the Colorado desert and presumably went home to their families after a busy day bombing, without ever facing any risk from their involvement. Not only was this asymmetrical warfare, but it was war without risk for those at the controls. The risk was solely on the ground where mistaken identities were rife, and wedding celebrations were confused for Taliban hotspots and bombed.

More recently in Gaza Israel has used various AI systems including Gospel and Lavender to determine targets. These systems determined targets for extermination with minimal human input. In the words of one Israeli intelligence officer. “I would invest 20 seconds for each target at this stage and do dozens of them every day. I had zero added value as a human… It saved a lot of time.”

A source who justified the use of Lavender to help identify low-ranking targets said that in wartime there’s no time to carefully go through the identification process with every target, and rather than invest manpower and time in a junior militant “you’re willing to take the margin of error of using artificial intelligence.”  This cannot be the future of AI.

E.O. Wilson warned that one of the reasons that we face a crisis is because we possess “Palaeolithic emotions, medieval institutions, and godlike technology.”

Let me finish with the words of Aran Levasseur:

“What makes a human life meaningful when machines can replicate or even surpass many of our abilities?

We have more freedom than any previous generation and less certainty about what to do with it. The result is visible: rising depression, loneliness, and anxiety in the most materially comfortable generation in history.

Prometheus stole fire from the gods and gave it to humans without the wisdom to govern it…We have always been better at expanding our powers than questioning whether we should. More sophisticated in our science than in our ethics. More capable in our technology than in our judgment. This asymmetry has never been more consequential. We are giving the next generation extraordinary power without the wisdom to wield it.

The most important question education can ask—the one Socrates never stopped asking—is also the simplest: What is the wise thing to do here?”

Tim Watson
Principal