This is the question I have. The answer - I'm still searching.
With this question I am like an insect looking up. The leaf looks like the sky, all there is. But the book Artificial intelligence, A guide for Thinking Humans by Melanie Mitchell seems the perfect place to start scratching the Unknown.
The scenario when your refrigerator or your entire house does not obey anymore, doing as it pleases, might be a pretty convincing movie scene about the future we are heading into, but for now it is still fictional. Or so I learn after having read this book.
Our devices, which we are using to manage life, work and transmit messages ( I intentionally do not call that communication) already know our deepest secrets but they do not yet act on that.
AI is a kind of “evil” technology, while it is emulating the way the human brain works it is also acquiring the power to behave like us.
The scary part when we speak about AI is that it seems to have agency, there is something humans don’t comprehend about it. It behaves as a force with will. There will come a moment when humans will probably lose their control over it. It’s not me saying that, but people who study and develop these things.
Melanie Mitchell is a professor of computer science at Portland State University and External Professor at the Santa Fe Institute. I guess this is why her way of explaining the mystery of AI is so understandable. She knows how algorithms are made, this is the technology constituting AI.
AI is learning in different ways but one thing is clear, scientists tried to copy the way the brain works and translate that in numbers, the language of computers. The mystery is not solved. There is something about the human brane that is not seen, not understood. There is a mystery to it, the scientists did not manage to explain it yet.
But can a being learn itself? Is the role of AI so that we could look at ourselves and thus, know ourselves? Is AI the result of human curiosity?
Multilayer neural networks are the foundation for modern AI. There are many human minds working on finding new ways to make these networks learn. But are computers actually are thinking? Do they understand? - yes. I did ask myself that too. At the moment this doesn't seem to be so. They are still learning, absorbing the reality we humans live in. The reality according to the human brain, the way it interprets it.
The hardest part for AI seems to be object recognition. The frenetic tagging which has been happening on popular social media platforms has pushed that to a superior level. AI learned mysteriously, from watching millions of YouTube clips, to recognise the cat but still the error rate is pretty high. Self driving cars rely on object recognition. But not each and every real scenario can be predicted and fed to AI. So this area is still blurry.
AI is learning exponentially, which is the basis on which scientists predict Human Level AI soon enough. AI does not forget and it does not make the same mistakes again. Compared to humans, who might be spending a whole lifetime doing the same things wondering why things don’t change, this is much better :) One example at this point is AlphaGo, the program which played against the (human) Korean Go champion Lee Sedol and won. Go is not a simple game. This documentary from the developers of AlphaGo , DeepMind, is mind-blowing.
I’ve learned about Deep learning, ConvNets, ImageNet, WordNet, Big data - AI needs something to munch on and we are all generously proving that- these are all ways humans are trying to use the power of AI in order to understand human nature.
Natural Language processing is another wide and thick topic. It is what AI does when translating from and to all kinds of language combinations. This possibility sheds light on some darkness helping humans be human, just that... AI does not understand, it has no sense of the meaning contained in those words. It is very interesting to find out how Google Translate actually works .
AI is fascinating and scary - its behavior is not predictable nor completely understood, it is complex and it is here to stay. Just like any other type of power, if given to the “bad” guy it might transform the world into hell.
Klara and the Sun by Kazuo Ishiguro - the soulfull product of a master human writer ( Nobel Prize in Literature).
Klara is the AI of that future we -the living- might not experience during this lifetime. Klara can do everything scientists are still trying to figure out at the moment. It moves, it can see, it can walk and talk, it responds and understands what is being asked.
In fact, Klara was created for the purpose of replacing Josie, the ill daughter of a mother who already lost a child. Losing another one would be too much for a human heart to carry. Klara was bought from a shop where she was sitting, watching the Sun, waiting for her turn.
Klara is observing the Sun. It is aware of its movements like no human is. Klara can recognise and enact emotions too.
That makes her special and so human-like. Klara is there to learn Josie. To become Josie after her death.
“Of course, the human heart is bound to be complex. But it must be limited. {...} Josie’s heart may well resemble a strange house with rooms inside rooms. But if this were the best way to save Josie, then I’ll do my utmost. And I believe there’s a good chance I’d be able to succeed” - said Klara, the AI.
I will not disclose the ending of the story- let’s keep the most human thing about humans alive - but I will leave you with one question : Do you think that the human heart can be imitated?
If Artificial Intelligence is having direct access to the human brain - and it does, influencing it in ways we do not understand, can it indeed become it? What is a human heart then?
What is human love? Where does it come from?
Masterful writers like Kazuo Ishiguro have this magical way of using words, showing perspectives to life we might not ever arrived to.
Not even AI.
Comments