The Red Shift...
...why the AI arms race may be tilted in our favour
Some of my favourite movies of all time take place in space; Star Wars (Episode IV, of course), Interstellar, the Martian, and Dune. I’m not sure exactly what it is about Denis Villeneuve’s remake of the Frank Herbert novel, but I can watch it on repeat, and I am literally counting the days until part II hits our screens early next year.
Since childhood, I have been fascinated by space. So much so that despite wise counsel to the contrary, I took physics A-Level, hoping that my average maths wouldn’t matter too much. (Ah, to be young and foolish…) I did at least manage to grasp some of the main concepts, and retain my fascination with physics, and a sense of wonder about deep space.
Recently, I have been thinking about the ‘Red Shift’, a concept which indicates that the Universe is expanding. In simple terms, light from distant galaxies is shifted towards the red end of the spectrum, suggesting those galaxies are both moving away from us, and accelerating as they do so. The Red Shift offers a helpful analogy as we think about the exponential technologies of the mid-21st century: artificial intelligence (AI), quantum computing, and synthetic biology.
These technologies are compounding. The more progress you make, the faster your rate of progress becomes. We see this in our daily lives as AI tools unlock productivity gains and inspire fresh technology breakthroughs in a range of disciplines, not least the two other exponential technologies. And just as distant galaxies move ever further and ever faster away from us, those societies which live at the cutting edge of AI are leaving the rest behind.
There has been a lot of debate recently about the risks to security posed by accelerating AI, and the threats which might arise from powerful models in the hands of our adversaries. At our recent symposium on AI + National Security during the UK’s AI Safety Summit, my colleague Kevin Allison’s panel explored these risks in detail. Much more work on this topic is needed, but I would like to make the case that, for now at least, the Red Shift is in our economic and national security interests.
Technological superiority is fundamental to economic and national security. This advantage, if we seize it, will be rendered structural by the Red Shift - we will move away, and we will accelerate away from our rivals. The West (and in particular the US) has, therefore, an opportunity to use its current position of dominance in all three of these exponential technologies to gain an unassailable advantage over its main adversaries, achieving permanent technology superiority, and thereby enhancing its security.
After a 25-year period in which China’s advance felt inevitable and geopolitical power moved slowly eastwards, the emergence of these ‘Red Shift’ technologies has changed the balance once again. China, the only real geopolitical challenger to the US, has made great strides, economically and scientifically, but it still has not matched the West’s ability to innovate, and with the dawn advanced artificial intelligence, based on access to massive amounts of data and compute, China may now be stymied.
It doesn’t matter if you’re only a little way behind - the Red Shift means you never catch up. Data and Compute are, truly, the essential resources of the mid-21st century, and while China’s use of the former is constrained by ideology, and the latter by export controls, it will be impossible to keep pace with the United States’ technological advance.
AI, quantum, and synthetic biology have the potential to deliver rapidly accelerating gains, each compounding the potential of the others and, to use another astrophysical analogy, allowing the most advanced among us to stay ahead of the event horizon, while the rest fall into a black hole. Welcome to the Red Shift.
Author’s Note: There is a debate raging about the wisdom, at a species level, of accelerating towards artificial general intelligence (AGI), which I do not attempt to address here. I am listening to and trying to learn from experts in the field on this issue. Nonetheless, the Red Shift inclines me towards the continuing pursuit of technological superiority for the preservation of our national security while we endeavour collectively and in parallel to understand and answer these existential questions. For a fascinating recent discussion on many of these issues, try this from the World Science Festival:
Authors note 2: I came across this post via Jeremy Howard (@jeremyphoward) on X which provides an interesting counterpoint to the argument set out above. I would be really interested to learn more about the compute and data resources which enabled development of these models:
https://t.co/5ABYoo7shU