“Human level AI”
For everyone of us interested in artificial intelligence, the idea of a human level AI that could understand and interact in the world like we do is charming and fascinating. For others, this is a really creepy idea. And everyone can understand why. But I am wondering what means “human level” AI ? What aspect(s) of our intelligence are we trying to artificially reproduce ? And when I think about intelligence, I also think about consciousness. Which is even more mysterious. What does that mean to be conscious and do we need to be conscious to be intelligent ? On a more technical level, AI today is mainly deep learning. And if we want to compare deep learning with human intelligence, with deep learning we already achieve better results (better accuracy) in specific tasks like image recognition. And it’s working because with deep learning we try to reproduce the huge amount of information processing happening in the brain that help us to make sense of new inputs, categorise and output a conclusion. But I personally don’t think about that when I think about intelligence.
I don’t want to offend anyone by giving my opinion on the matter of our intelligence and consciousness because no one really knows. But I don’t believe intelligence is happening in the brain. Or not only. And I don’t believe in magic neither. I mean that if there is magic in the universe, I don’t think it’s here. I am more comfortable thinking that the brain is the decision maker responding to… well a lot of stuff. let say our intelligence is the connections between evolution, experience, senses, memory, emotions, hormones, our capacity to analyse our environment and take decisions and our ability to make connections between new informations and past experience and to empathise, adapt, communicate and find our place in a society. And a lot more. Consciousness would be a more abstract and philosophical object that could be knowing (or at least feeling) that we are “intelligent”. And another important aspect of intelligence would be the capacity to invent and imagine.
Humans are complex organisms with the only goal to reproduce. This creates competition and needs. We also direct a lot of our decision making and “free will” toward challenging our inevitable death and our desire of sex without procreation. Another really important thing is that we take a lot of irrational decisions. Our emotions play an important role in our decision making process. Adding bias by creating an opinion or even making totally irrational choices. We do stupid, and sometimes dangerous things for love for example. And I don’t think we can fully understand our intelligence without really understand what love is to us. Because I believe those natural needs and emotions shape our intelligence and decisions landscape. If we don’t artificially reproduce the machinery of human body, natural needs and body / hormones cycles, human level AI is — in my opinion — not possible.
Being aware of our own mortality has also contributed to the expansion of our intelligence, we created myths, religions and sciences, we started to survive and imagine things we can’t even see. We created art. We became anxious, philosophical animals. We became jealous, envious and strong enough to pursue more and more. We created numbers. And our love for numbers (amount of money, grades, the size of our house…) and awareness of mortality drive us to destroy our own planet and to kill each other. We are also proud of being human. Because we don’t want to believe in other biological organisms being as smart (whatever “smart” can be) or smarter than us. At least not on this planet. We believe in a human ethic. We decided our lives should have more importance than any other species lives.
If we are afraid of general human level AI, it’s because we are afraid to lose what makes us ‘human’. We like the mystery behind our conscious / intelligence. What if we realise our intelligence has nothing mysterious, but is just the result of electrical impulses through a stack of layers of neurones. That there is actually nothing special about being human. What if we realise we already are ‘artificial’ reproductions of another type of intelligence…
I like to think that our irrationality and passion, as well as our anxiety of mortality and loneliness is what makes us human. I like to think that that’s what drive us to create amazing and beautiful things, to write heart-braking stories or paint true masterpieces. To laugh, to fall in love, to translate smells and colours into memories and write beautiful songs. As well as our ability to appreciate all of that, because we can empathise. We can feel someone else’s emotions.
The biggest (and weird) source of knowledge about our world and ourself today is our data (thanks to numbers and science). We know everything about everyone, and we realise that we reproduce patterns. Again and again. We realise how biased our decisions and opinions can be. That if we want to know something about ourself, “Data” knows better. And instead of being afraid of it we should take really care of it. So data are facts. And we create LOADS of data.
So what if instead of dreaming of a human level AI, we dream about a ‘conscious’ digital complex organism that would take reasonable decisions based on data, distribute energy, resources, knowledge, free humans from stupid and repetitive labour, create a new economy…
We could redefine what makes us humans as individuals and as a species. We could redefine our needs, rethink our way of living, think about our expending appetite for numbers. We could start connecting again with each other. We could get back the joy of observing and creating beautiful things, to share and love. We could create an AI that’s not a ‘human level AI’, but a global artificial conscious. Aware of its data. That will not become more ‘human’ but will certainly prevent real humans to become more robots and actually less ‘human’. This is at least the AI I would be excited for.