Alibaba chairman Joe Tsai is out there dropping truth bombs about AI and robotics
Real talk: Joe Tsai knows what’s up.
Joe Tsai, the chairman of Alibaba, a big tech company based in China, recently took the stage at a speaking gig aimed at young entrepreneurs in Hong Kong. In a surprise twist, he spent a significant portion of his time criticizing the AGI and robotics industries. This, despite the fact that Alibaba is a global leader in both.
Some people just like to watch the world burn.
All jokes aside, we are 100% behind Chairman Tsai. According to him:
AI can’t reach “human level” intelligence without “emotional intelligence.”
Humanoid robots don’t make sense for the purpose of doing useful work.
Here’s the exact quotes we’re referring to:
“All the positive encounters that you have with people that you love, people that you really enjoy spending time with, that judgment, that data, how does the machine capture that data in order to train AI to replicate the positive energy?... I am not sure how we train the machine(s) to make them ‘smart humans’ or ‘smarter than human beings’ with EQ and compassion.”
“A lot of what these robotics companies are trying to do is to replicate a human being that moves, feels, walks, runs and talks and thinks like a human being. The question I want to ask is what is the application of a humanoid?”
Emotional Intelligence
In the first quote, he’s referring to emotional intelligence, one of many types of intelligence that philosophers, psychologists, and other experts on intelligence have identified.
Is he right? Are the smart people at all the smart-people institutions right? Nobody knows. We don’t have a ground-truth explanation for intelligence, consciousness, thinking, reasoning, etcetera.
Humanity is trying to figure all this stuff out in real time. Here at the Center, we really like this paper by Lab42’s Rolf Pfister, called “A Representationalist, Functionalist and Naturalistic Conception of Intelligence as a Foundation for AGI.”
Humanoid robots
On the subject of humanoid robots — robots that intentionally resemble humans — Tsai expressed that he didn’t think they made sense. He pointed out that industrial robots aren’t designed to look like humans because they’re made to do a task.
The bigger picture here is that humanoid robotics proponents such as Elon Musk claim that robots should look like us so that they’ll be well suited to doing the tasks we’re already doing. The thinking here seems to be that if you give a robot a hand, it can use a screwdriver, a hammer, or a dishrag.
We think that’s stupid.
It’s our opinion that the only reason to make a robot look human is for aesthetic purposes. For the same reason you wouldn’t make a cockroach your restaurant’s mascot, big tech knows it needs to design robots that look friendly and cute.
Of all the possible designs for a useful robot, humanoid doesn’t seem anywhere near the top of the list. A trashcan on treads with a flat top and a modular, multi-tool, extending limb design would probably be much more useful than a robot that has to put down what it’s carrying every time it changes tasks.
And, without the aforementioned EQ, a robot that tries to look and act human is, and always will be, creepy. (See: the Uncanny Valley)
Tsai also questions what the point of gendering machines is; he appeared to wonder whether there was some way to involve biological processes such as hormones in the development of “human-level” artificial intelligence.
We’re not sure what the point of gendered robots is either, unless they’re meant to be used for entertainment purposes.
Read more: What is artificial general intelligence? — Center for AGI Investigations
Art by Nicole Greene
