The Answer Is Fairness. But What's the Question?
I think the answer would be fairness. But what’s the question?
If you asked most people in the UK to agree on something I think you could get them to agree that we should offer everyone a fair shot.
But if things don’t feel very fair right now I can assure you, AI is set to make that worse.
Why?
Well, do you think we should all pay the same amount for goods and services?
We won't if algorithmic pricing uses our location data and browsing history to set prices?
Have you met someone who lost their job to an AI?
If not, it’s only a matter of time.
Have you tried to appeal a decision made about you by an algorithm — a loan rejection, a CV screening, a benefits assessment — and found there's no human to talk to?
Over the last few months I’ve become more and more convinced that putting fairness back at the heart of things is an important route out of our current mess.
This is going to be especially important when it comes to how we distribute the spoils of AI.
Putting to one side the looming bubble (which risks making us think AI doesn’t matter so much after all - false comfort IMO), take a minute to think through the basic equation of AI when it comes to value.
It has hoovered up pretty much the entire internet (which itself hoovered up most of history to the turn of the century) for free, and faces very little serious pushback from politicians. Not exactly fair.
We now face new models, AI agents and unpredictable consequences throughout our lives with no meaningful regulatory safeguards. Also not fair.
The capabilities of AI models keep evolving. Loads of the early innovative start-up ideas people have had are swallowed by the model behemoth which simply spots a good idea and absorbs it into the ever-growing wheelhouse of things an AI can do itself.
At the same time, the depth of our data-based relationships with AI will keep growing. How long before it’s pretty normal for an AI to read everything in your phone, your inbox and your calendar in order to offer you tailored advice?
Every bit of data we give away for free creates value for AI companies.
Every bit of work we outsource to AI creates value for AI companies.
Every piece of creative work scraped without permission makes the model more capable and the original creator less necessary.
Every bit of AI’s massive forecast computing needs will require state infrastructure and taxpayer investment.
Every bit of AI’s massive water and energy needs makes managing the climate transition harder.
In return so far we get a messed up information environment, fewer jobs for graduates and a stuck needle on productivity.
If you think things feel unfair now, give it another decade of unregulated AI.