Desirable Distractions
A thought that popped during a long drive to home
The traffic jam
Rohit, my colleague, and I left office at 7.20PM.
I felt triumphant after sending out a stinker to the motor service POC’s manager. There was a sense of accomplishment in believing that my car would be back in a couple of days thanks to the intervention of the general manager of the service station.
However, this happiness was short lived. We were stuck in a bad traffic jam. A drive that generally takes 30-45 minutes was going to consume more than an hour today. I thought it was a good idea to play an interesting podcast to keep both of us engaged during this ordeal.
So I started a Lex Fridman episode. Both of us ended up learning a lot, discussed concepts and explained the contents of it to each other when either of us lost track of the dense conversation.
Summary of the podcast (stuff I remember)
It was a highly engaging podcast about AI, reinforcement learning, natural and biological neural networks, supervised and unsupervised learning, nature of learning and the future of artificial general intelligence. Some random notes I made are below:
Do we need to know how the brain works to create a machine that thinks better than ours? We were able to build a plan without knowing too much about how birds fly.
Differences between artificial and natural neural networks : need for huge amount of labelled examples. Humans don’t need this humungous labelled data to learn.
Will monkeys be able to identify faces if you hide all faces from them during the first few weeks of their lives? Apparently no. This was discovered through a research study.
Understanding the cortex could be key to solve problems related to speech and vision. The same hardware handles both vision and speech. But in the domain of AI both these problems need different approaches/hardware to work.
Why do we sleep? Apparently there’s no substantial answer to this question.
Does adding more parameters to a model and reduce the data fed to it lead to a more reliable model than one with fewer parameters and more data? We don’t know.
One can draw clear distinction between the hardware and software of a computer. The guy designing the transistors need not know about the powerpoint software driving the computer. But that is not the case for human brain. In that sense god is a software and a hardware guy.
Can consciousness be fed into a computer? Does human consciousness exist without the blood and flesh? Apparently we can simulate consciousness in the computer, but we can’t be sure whether it can mimic human consciousness.
Can we understand the logic behind ethics and areas of the brain which control it? Is it possible to simulate ethical behaviour in a computer the way humans can be taught ethics? Probably yes.
Once you design a learning machine, you can’t control what and how it learns just the way you can’t control what kind of music a baby might like or dislike. It is hard to understand how exactly things will turn out.
You might understand how a hydrogen bomb works. It is possible to go into as many details as possible, but the true micro-phenomenon is hard to understand.
Is it time to worry about the AI doomsday? It’s good problem to worry about, but it sorta too soon. Maybe we might face some sort of challenge later, but not in the near future.
Desirable Distractions
If not for the podcast, I would be really frustrated about the traffic jam. This ‘brainy’ distraction helped Rohit more than me because he gets agitated by bad drivers on the road. He admitted to this reality when I was about to get down from his car.
I left him with one of my ‘philosophical’ lines. I confessed that this is the way I distract myself from all the bullshit in my life. Which is kinda true.
How long can I keep on thinking about the long line of ladies who destroyed my heart in dozens of different ways? How many times can you think about the guy who pissed you off at work or the nasty remark made by your mom in 2014?
You have to distract yourself with something more beautiful, intriguing and interesting. Something that makes life more exciting. That has been my funda for many years now. I’ve probably done this unconsciously. But this evening made it more visible and obvious.
And I’m glad that my mind has chosen this path to cope with all the bullshit thrown at me. This approach is inadvertently making me better in some way or the other every single day.
P.S. I bribed myself with a cappuccino and a croissant to motivate my lazy ass to sit and write this post after a long day.