To become truly intelligent,
machines depend on human ingenuity

Lesen Sie die deutsche Version hier.

In 1863, Samuel Butler, having fled industrial England for a sheep farm in New Zealand, warned the public – for probably the first time in history – of the risks to the human race of mechanical life. In a satirical letter to the editor titled «Darwin Among the Machines», Butler foresaw the creep of sophistication among machines that mirrored the biological adaptations Darwin saw in South America four years earlier: only a few millennia passed from when the lever was the most intricate tool known to mankind to the invention of the steam engine. He prophesied that «we are ourselves creating our own successors». The machines shall be improved and improved until «we find ourselves the inferior race». Before the computer was ever invented, Butler foreboded that «man will have become to the machine what the horse and the dog are to man». Just as the domesticated dog is better fed and better cared for than its feral ancestors, mankind will undoubtedly improve its material condition as the march of machine intelligence continues, but there will come a point when man is no longer the apex species driving the planet’s destiny.

The only remaining question, which is still the one we have today, is one of time: When will the machines become supreme? For all prior centuries, a so-called technological «singularity» was simply impossible. Butler had no risk of being overthrown by a steam engine, but the evolutionary rise of artificial intelligence in our electronic age may be our undoing. Humankind has shown that the supremacy of the planet rests not in physical might, but rather in inventiveness. We achieved dominance on this planet not by wrestling gorillas, but by inventing the bow and arrow. Whereas in Butler’s time only our physical strength was threatened by the machine, now, our intelligence – the very facet of our species that enabled our rise – is being threatened to be overtaken by the machines. Nevertheless, even though machines can compute algebra faster than humans, there are few today who doubt we can still simply pull the plug on any existing electronic machine and regain control.

The hardware is not the problem anymore

Until roughly the past twenty years, we simply lacked hardware powerful enough to rival the human brain. Now, however, our largest supercomputers exceed the complexity of the human brain. Now that we have the hardware capable of running such complexity, we are merely missing the right theory, or software, to generate an artificial general intelligence. We now enter the dangerous period of history where the singularity could realistically happen at any moment when a spark of genius erupts in some laboratory. Although unpredictable when it may occur, it is now purely a matter of time.

Some advocate that the recent advancements and emergent intelligence in large language models will, as they continue to scale up, achieve super-intelligence. Yet, these models, despite their raw powers of computation, lack any true sense of agency. No matter how sophisticated or knowledgeable these models become, predicting the next token of text can most likely never emerge a conscious being that can rival human power. It is not raw intelligence that enabled the rise of humankind, but the conscious application of our knowledge to the environment. Scale is not all you need; you need intentionality.

«It is not raw intelligence that enabled the rise of humankind, but the conscious application of our knowledge to the environment. Scale is not all you need; you need intentionality.»

Can computers be conscious like humans?

Consciousness is the missing ingredient. For the machines to realize their full potential and Butler’s vision, they will need to learn not just to think and compute but also feel. Some scholars claim that consciousness is an evolutionary appendage like the unnecessary appendix. Philosopher David Chalmers popularized this theory when he coined the term of a «philosophical zombie»: a being that in all aspects resembles the intelligence, emotional affections, and personality of a human, but lacks any inner conscious monologue. A philosophical zombie can perfectly mimic outward human behavior but have no interior experience. The fact that we can fathom the existence of such a being shows, Chalmers argues, that consciousness is extraneous to intelligence. Yet, our bodies and brains expend untold amounts of energy on our conscious experiences, suggesting that high-level intelligence is interlinked with consciousness. Otherwise, evolution would have long ago evolved a more energy-efficient unconscious intelligence.

Other critics suggest that computers can never be conscious like humans. There is, they stipulate, something uniquely biological in the inner workings of consciousness. Ever the evolutionist, Samuel Butler rebutted this rebuke in his 1863 letter: «(assuming the theory of evolution, but at the same time denying the consciousness of vegetable and crystalline action) the race of man has descended from things which had no consciousness at all… there is no a priori improbability in the descent of conscious machines from those which now exist». While we may not know the ingredients necessary to produce consciousness in a machine, that is little comfort that we will not one day discover it.

Only time will tell when humanity will pass the baton to the artificial. It is not, however, a bet in the sophistication or correctness of the computers or the algorithms we have today, but rather in the genius of humans working together to design solutions. For, as Butler recognizes, «man at present believes his interest lies in that direction; spends an incalculable amount of labour and time and thought in making machines breed always better and better … he has already succeeded in effecting much that one time appeared impossible». Faith in the rise of the intelligent machine is not faith in the artificial, but rather a belief in the ingenuity of the smartest and brightest humans to materialize its existence.