Debate Class 2/22
2023 Debate Competition
The development of artificial intelligence benefits humanity
Practice
Topic: Resolved UCLA is better than Harvard
Framework: We believe that a school which can make students feel happier is a better school.
Claim one: UCLA has a higher educational reputation.
Claim: UCLA has better weather.
Rebuttal: it is totally not true
Prove it with evidence.
There's no supporting evidence
A student is not an expert, so the evidence is not solid.
A student says that UCLA never snows
Organization
Let the judge understand which point you are attacking at
Try to organise your rebuttal against the previous case.
What is Artificial Intelligence
There is no consensus definition of AI
The Hard Problem of Consciousness is a term coined by philosopher David Chalmers and is the starting point for understanding what is AI
Most definitions of Ai are about replicating the behavior of humans with machines
If we don't understand how human thinking works in the brain, we cannot recreate it in a machine
The brain is an incredibly efficient machine. We need whole buildings and Megawatts of electricity to recreate what your 1.5kg brain does with about 20 watts
Alan Turing is sometimes called the father of computer science after inventing a computer to decrypt German coded messages in WW2
He wrote on 'can computers think?' he proposed the turning test where humans text with people and chatbots, and if they can't tell the difference than the computer is "intelligent".
Stuart Russel is a brain surgeon and wrote a modern textbook on AI.
IBM describes AI using Russel's 4 definitions
Human Approach:
Systems that think like humans (process)
Systems that act like Humans (results) "Turing Test"
Rational approach:
Systems that think rationally (process)
Systems that act rationally (results)
Process vs Results
Process is about the method with which a system makes a decision. This process could be try to simulate humans or could just follow rational guidelines.
Results focuses on whether it seems human/rational when its done.
Human like vs Rational
We could try to replicate humans' thinking process or we could try to get results that are equivalent.
We could focus on rational decision making instead of human decision making if we think we can make a superior system that way.
All 4 ways of approaching AI are valid
The turing test mentioned before is concerned with results rather than how those results were achieved. The goal is to appear human like.
Human like process
Deep learning or machine learning is where a computer develops its own algorithms to try to explain patterns in data. It simulates the way humans learn through experience.
We often don't know the rules of the algorithm the machine is using but through trial and error of emergent behavior, it works.
Rational process
Games like Tetris are simple enough math that a computer can be programmed the ideal way to solve the math problem.
It's difficult to write these ideal logic rules. Chess is only solved with seven or fewer pieces on the board. Eight is too difficult, too many options.
Rational Results
One desirable trait of AI would be if it could handle a novel situation, one which it has not already been preprogrammed for but still deliver desirable results.
The problem with this is sometimes there is no perfect outcome.
Core conflicts with developing AI
Human Obsolescence
Shifting Economic Power
Robot Rights
If people are going to be obsolete in the future, what are you learning for?
Students are worried school is wasted if they can never actually compete with computers by the time they become an adult.
Generally this creates 2 crises: an economic crisis for workers and a philosophical crisis of purpose for humanity.
The creators of Ai and robots are effectively replacing labor with capital. Why keep workers if robots can work 24 hours without food, homes, etc.?
What this functionally does is shift more economic power from workers who make money from spending their own time and energy, to the rich who own the companies and now the robots.
For hundreds of years, the majority of people have worked to survive.
Many sci-fi books imagine the near future as a disaster with mass riots triggered by mass unemployment due to automation.
Society could just try to share resources, but its usually assumed it will take violence to make it happen.
Even if its not that bad, we do see surging inequality globally, and all the resulting harms of inequality.
Unethical uses
If Ai replaces humans it will make large contributions to every industry and human experience.
Not all of these contributions seem to be ethical or positive. AI can do both the positive things better AND the negative things better. Ex: Racial bias in data.