After a suspended Google engineer claimed that the company’s artificial intelligence (AI) system called LaMDA had developed sentience, a public debate about the possibilities and limits of AI ensued. What are the facts? Are humans doomed to be ruled by ultra-smart silicon overlords?
Ever since humans built the first machine, people have speculated about how the mind works using the most advanced machine as an analogy. At one point, philosophers thought that consciousness worked like a windmill, then like a steam engine, then a telephone exchange. Today, it is widely believed that our mind is just software running on a biological computer – the brain.
However, this is built on a philosophy called materialism – the idea that consciousness is an illusion, that only matter exists. If we could map every neuron and chemical in our brains, our behaviors could be modeled perfectly on a sufficiently advanced machine. It may seem like we have a consciousness and free will, but materialism says that is just a mirage. You are your atoms and nothing more. Therefore, when the former Google engineer claimed that LaMDA has become sentient, he just proclaimed that it now is sufficiently advanced that it behaves indistinguishably from a human.
But is it true? Are we just advanced toasters? Evolution proves that consciousness is a real phenomenon, not merely an illusion. Nature is full of biological variation; some are better at survival than others. Natural selection acts like a filter that removes the unfit from long-term cross-generational survival. Two criteria must be fulfilled for natural selection to work: The trait must be genetically heritable, and it must be real.
Natural selection does not work on illusions. There is no reason natural selection would favor the highly structured inner experience of consciousness if it were only there as an impotent ornament. Thus, consciousness likely evolved hundreds of millions of years ago in primitive organisms and has been refined and structured into greater complexity.
It implies that consciousness is not merely “software,” but a natural phenomenon generated by a precise physical process. The famed physicist Sir Roger Penrose believes it has something to do with quantum mechanics and has identified microtubules in the brain as the likely structure that produces it.
Or maybe consciousness is summoned? Biologist Rupert Sheldrake thinks so. He claims that the evidence shows that the brain acts more like an antenna that receives and amplifies consciousness than a machine that generates it.
If Penrose or Sheldrake is correct, the chance of a modern-day silicon computer ever becoming conscious is slim. Computers were made for a completely different purpose, and the likelihood that it by extraordinary coincidence also is the correct way to produce sentience is vanishingly tiny. LaMDA is still just a toaster, albeit a highly advanced one.
Therefore, we can all sleep safely knowing that AI silicon overlords will not take over the world. A far greater danger is that AI will be the tools of organic sentient Silicon Valley overlords to surveil, data mine, censor, and control us, using their wealth, corporate and technological power.