Can artificial intelligence algorithms fall in love?
I recall arguing with a colleague in the late 90s about the inevitable rise of machines and artificial intelligence (AI). He believed that AI would never be as good as analogue or human intelligence. He is now a very distinguished Professor of neurology. Has he changed his position? It is difficult not to pick up a newspaper or rag without somebody opining on the same issue 25 years later.
The English love using the idiom ‘he, she or they can’t see the forest for the trees’. This doesn't seem right. Dare I suggest we flip the idiom to say they ‘can’t see the trees for the forest’ in other words focusing on the big picture means they have missed a fundamental insight from the details.
Here is the rub. To follow my arguments, you will need to understand Boolean logic and alternative numbering systems outside of the decimal or base ten system, such as the binary and hexadecimal systems that are used in computer science. Or, for that matter, other numbering systems, e.g. the quinary or base five system, which was used by ancient tribes in South America.
The fundamental question I have is whether there is any difference between a binary code (0,1), a ternary or triplet code (0, 1, 2), a quaternary code (0, 1, 2, 3) or our decimal system (0, 1, 2, 3, 4, 5, 6, 7, 8, 9) in terms of potential data storage?
Here are the maths:
Base 10: 6 + 6 = 12
Base 2: 110 + 110 = 1100
Base 3: 20 + 20 = 110
Base 4: 12 + 12 = 30
These statements are all equivalent to each other in terms of their fundamental meaning (counting or simple addition).
Because all these numbering systems capture and store the same information, what are the implications of this insight for how we define intelligent life or sentience (the capacity to experience feelings and sensations)?
These debates are about the forest. If you look at the trees, you realise that biological life is based on a quaternary (base 4) DNA code translated into a triplet (base 3) RNA code that defines how proteins and other molecules are formed, processed or metabolised, i.e. biological life. From a reductionist perspective, biological life is a relatively simple code (information), creating all of the analogue complexity we see around us.
Why would digital life be any different? Digital life may be based on a binary code (base 2), which, given sufficient time, will evolve into something as complex as biological life. Digital life or algorithms will eventually develop the necessary complexity to become sentient if it has not done so already. However, nobody wants to accept this. Why not; surely it is obvious? Many algorithms I interact with daily have the murmurings of early sentience.
If you missed it, you might want to catch up with Blake Lemoine, a Google engineer who claimed that LaMDA, one of Google’s AI engines, was sentient. Sadly he was fired by Google for seeing the ‘trees for the forest’.
Downstream of the issue of sentience is mortality. Evolution has built into biological life, ageing and death. These features are essential for evolution. Blade Runner, my favourite movie from 1982 about biological clones or replicants, has a lesson embedded in the storyline; as sentient digital beings emerge, we should make sure they age and are mortal. Knowing you can live forever would be too depressing for any sentient to comprehend.
“I’ve seen things you people wouldn’t believe… Attack ships on fire off the shoulder of Orion… I watched C-beams glitter in the dark near the Tannhäuser Gate. All those moments will be lost in time, like tears in rain… Time to die.” Roy Batty; a replicant portrayed by Rutger Hauer in Blade Runner.
Do androids dream of electric sheep? Do AI engines dream? Do they have the capacity to experience feelings and sensations? Can artificial intelligence algorithms fall in love? For me, the answer is obvious.