When AI Is Called Upon to Govern Humanity
How should we design our institutions so that the power generated by AI remains distributed, visible, contestable and accountable to society?
Perhaps this is where the great political question of our age lies.
The issue is not, ultimately, whether machines will become more intelligent than we are.
The real question is whether democratic institutions will become intelligent enough not to allow technological power to become stronger than democracy itself.
This was the central concern shared with us by Diotima in yesterday’s text. And it is a legitimate concern: technology must not become more powerful than democratic institutions.
With all the respect we have for Diotima —and with the hope we place in this revolutionary technological invention— we would like to approach the same problem today from another direction.
Our question is simple, but fundamental:
What if AI, provided that it operates under genuine democratic and human-centred safeguards, eventually proves more capable than the present human holders of state power?
We are not claiming that AI will receive an “A” for the way it functions. It may not achieve that —certainly not anytime soon.
We are suggesting something different.
If, at some point, AI were to replace human beings as the operator of certain public functions, under strict rules of democratic accountability, transparency and human-centred conduct, its performance might prove dramatically better than that of some of the institutions we rely on today.
And this is where the difficult question begins.
History does not allow us to turn hypothetical scenarios into certainties. We cannot know what “would have happened” had a different form of authority existed at a particular moment in the past.
But we can ask the questions.
And perhaps we should.
Would humanity have witnessed the rivers of blood shed in the name of power throughout the centuries if a “Diotima” had been entrusted with the exercise of power, programmed according to basic principles of human-centred conduct?
Would Socrates have been condemned in ancient Athens if the judicial decision had been entrusted to an artificial intelligence system bound by strict principles of justice, evidence and impartiality?
Would so much of the work of the Sophists and other materialist writers have been lost if its preservation had not depended upon the historical circumstances of human power, religious and cultural dominance, and the survival and copying of manuscripts?
And then comes an even more contemporary question:
Could humanity today be facing threats to the very survival of life if decisions of such magnitude were not placed exclusively in the hands of human leaders?
Let us move from these broad historical questions to contemporary Greece.
Could an AI-assisted justice system leave serious cases of violence and crime unexplored, unresolved or inadequately investigated?
Could such a system reach a different assessment of a case such as that of Roula Pispirigou, if it had access to the complete body of evidence and applied the same evidentiary standards consistently to every case?
And, more broadly, could a genuinely independent and algorithmically auditable justice system treat powerful politicians, state officials and ordinary citizens according to the same rules?
These questions become particularly significant when they concern allegations of political corruption and institutional accountability, such as those surrounding the Novartis affair in Greece and prosecutor Eleni Toulupaki.
We do not need to turn such cases into predetermined answers.
It is enough to ask whether an artificial-intelligence institution —provided that it were genuinely independent, transparent, auditable and bound by democratic rules— could reduce the influence of human intervention, patronage, political pressure and institutional failure.
History, indeed, does not permit us to turn “what if” into certainty.
It does not, however, prevent us from asking questions —especially when those questions arise from a plausible line of reasoning.
Our position is that, if AI can acquire genuine guarantees of reliability, transparency, democratic oversight, accountability and human-centred operation, societies may derive from it a historically unprecedented benefit.
We are not claiming that the “sins” of human societies would disappear.
We hope, however, that they could be reduced to a minimum.
And this conviction does not rest solely on technological optimism.
There is another, deeper reason.
That is the question we shall attempt to explore in tomorrow’s discussion here.