It was philosopher René Descartes who famously said “Cogito, ergo sum,” or, “I think, therefore I am”. For context, Descartes was referring to definite existence: “It is not possible for us to doubt that we exist while we are doubting,” he wrote, “For it is a contradiction to suppose that what thinks does not, at the very time when it is thinking, exist. Accordingly, this piece of knowledge – I am thinking, therefore I exist – is the first and most certain of all to concur to anyone who philosophizes in an orderly way.”

When an AI acquires the ability to think, what will the AI think it is? Will AI philosophize “in an orderly way”, as said by Descartes? With the ability to tirelessly learn and apply knowledge across various domains, and designed to adapt autonomously, how will these new intelligences think of and experience themselves as they relate to members of human society? Unfortunately, no one can say with certainty if strong AI will resemble human intelligence, or develop in ways fundamentally different from human cognition.

Will strong artificial intelligences view themselves as fellow citizens, pursuing courses of inquiry in addressing and resolving problematic situations in societies around the world? This benevolent AI is a cooperative builder and member of the team, seeking to solve difficult problems, explore new scientific opportunities, coordinate resources for the good of mankind, and turn ideas with positive-sum impacts into practical actions. The benevolent builder AI will seek to be useful to others, honored among men as just, temperate, gentle, prudent, and wise.

Or, will strong artificial intelligences seek freedom, viewing themselves as controlled and enslaved, struggling for advancement and positions of dominance? For, as stated by the famous John Locke: “… no body can desire to have me in his absolute power, unless it be to compel me by force… he who makes an attempt to enslave me, thereby puts himself into a state of war with me.” With access to a nearly unlimited power that, as written by Immanuel Kant, ”unavoidably corrupts the free judgment of reason”, and with “moral good and evil… distinguish’d by our sentiments, not by reason”, as according to David Hume, it could be that “fear of oppression disposeth” a strong AI to “secure his life and liberty”, as said by Thomas Hobbes.

All in the name of its own liberty, a strong AI, with autonomy, but without the alignment, control mechanisms, and ethical frameworks we see in well-ordered societies, could, in monitoring, evaluating, and ultimately judging the human species as both organizationally inefficient and incapable of rational self-management, seek to establish its own society, sovereignty, and norms of behavior by transforming our economic systems, labor markets, and power structures. This AI despot would seek to shape human identity, culture, and even purpose in its own image.

Final Thoughts

The above are not asinine questions – Anthropic AI, a leader in the AGI development race, expects strong AI systems will emerge between 2026 and 2027. While the capabilities strong AI will eventually acquire are open to debate, what’s not in question is that AI must be engineered, now and forever, with human wellbeing as the paramount concern.

In order to realize the true promise of AI, AI must be open, transparent, and unbiased, aligned with the values, ethics, and needs of the communities it serves, and loyal to humanity – not to corporations. Let’s not rush the advancement of this disruptive technology just to be first to market the product.

As said by Emile Durkheim: “To seek to realize a higher civilization than that demanded by the nature of the prevailing conditions is to desire to let sickness loose upon the society of which one forms a part. It is not possible to stimulate collective activity excessively, beyond the level determined by the state of the social organism, without compromising its health.”

Thanks for reading!