The Sylvan artificial intelligence debate considered the following motion:
This house believes artificial intelligence will benefit humanity.
The debate took place on Monday, 5th June. Katarina Pera proposed the motion and Tony Koutsoumbos opposed it.
The proposition arguing that artificial intelligence will benefit humanity
The proposer began by referencing a New York Times article from 1894 decrying a threat to women’s health (the bicycle). We cannot throw the baby out with the bathwater every time something new comes along. AI has many positive uses such as detecting early-stage cancer and treating genetic diseases. We now take translation apps for granted. What about jobs? AI can replace unsafe ones with safe ones. More people can get into IT because it democratises coding, though even those jobs will go away. If we don’t drive output we can’t eat, driven by capitalism. Yet we are still worthy of shelter, food and love.
The proposer continued
Did painting go the way of the dinosaurs after photography? No, new styles emerged. AI-enabled art can support self expression. When AI incorporates existing art, the artist should get royalties like musicians do. We should not discount AI due to humps in the road, which we can resolve with laws and code. We need to stretch our minds.
The opposition against the artificial intelligence debate motion
The opposer pointed out that AI usually swings between utopia and the apocalypse, while the proposer gave a realistic take. We cannot make accurate predictions of the costs and benefits of AI in the long term. Yet we can draw on past and present impacts. I won’t say the Terminator is real. Progress comes at a cost, such as AI expanding in warfare – we don’t want to develop certain weapons. We can’t say with certainty that we will benefit. We need to prevent potential harms, yet we can’t say we are taking enough action, and who is responsible? Cambridge Analytica undermined the integrity of our democratic systems. Chat GPT has ethical guards but they have limits; driverless cars have led to deaths. Tech giants won’t disclose their algorithms. US parole decisions show that AI incorporated biases against blacks.
The opposer continued
Today’s AI started in 2000, we’ve had Facebook for 20 years, and no one anticipated the storming of the US Capitol. We have a track record of overestimating the benefits. Only Italy has focused on taking on the challenge. What might be the consequences of getting AI controls wrong? If we had to weigh the harms of the past that we didn’t see coming relative to overstated benefits, is it worth it?
Floor speeches from the audience of the artificial intelligence debate
Floor speeches covered the topic from a wide range of perspectives. AI has led to echo chambers of customised social media feeds, each with their own reality and no opposing views. This creates a divisive society. AI algorithms now can outthink humans, and as black boxes we can’t understand drive an extinction risk. How will it be regulated? Rather than the tech, social policy holds the key to solving this issue. Tech has always interacted with labour markets, and even in the 1930s Keynes predicted the 10-20 hour work week. Microsoft claimed its operating system would reduce work! We need to get control out of the hands of the tech barons. Cambridge Analytica led to Brexit and Trump at the margins. We need active government to share the benefits with workers. Our data feed the models, we have the democratic right to regulate them.
Floor speeches continued
The models suffer from the ‘garbage-in, garbage out’ problem. Yet the data come from wealthy countries, providing another disadvantage for developing countries. Pupils write papers via Chat GPT. General purpose AI could become sentient, whereas specialised AI is already the reality and we will use it. Condensing the knowledge of thousands of people provides instrumental value. AI already damages humanity by allowing us to avoid thinking about what we do. Our trains run safely based on human-coded software. AI may not let us turn it off. Applied to climate change, it could decide that removing humans will solve the issue. Yet the data we put in represent us – is humanity good or evil? Throughout history we have killed each other and rigged elections. Without technology, we would mostly be peasants farming the land.
More floor speeches
I just ran the motion through Chat GPT – the answer has four paragraphs on benefits and one small one on potential downsides! We need to understand whether AI is fundamentally a weapon – is it a new Kalashnikov, a more efficient killing machine, or nuclear fission, something that could kill us but we’ve managed to control? AI will make things safer – humans won’t be allowed to drive. Will we trust AI more than our own opinion?
The opposer’s rebuttal
In rebuttal, the opposer raised the spectre of programmes that can teach themselves. The question rests on overestimated benefits versus underestimated harms. Predictions have AI increasing GDP by trillions of dollars. Driverless cars, you don’t hear about them anymore, many went bankrupt. We have normalised echo chambers. AI will replace entire industries, not just jobs. The industry itself has warned of the existential threat to humanity. Will government come to the rescue? Should we change the data inputs? Or become a fly on the windshield?
The proposer’s closing speech
In closing, the proposer, as a data scientist, pointed out that it takes hard work to build AI. We can break down the black boxes, we know the input data. We have agency and the power to regulate. Autonomous cars haven’t yet launched due to regulation, we got seatbelts in the 1970s. GDPR shows regulators can act. Deepfakes are threatening and scary. The social policy point was correct. Anti-black bias in parole decisions came from data from biased people – it brought the problem to light and we can quantify the bias. We can spot problems and work on eradicating them. We need frameworks for control over data, to remove bias and privacy rules. AI can help prevent world hunger and predict earthquakes, but we need to use it carefully.
Result: the artificial intelligence debate motion carried
In the final vote, the Sylvans concluded through the artificial intelligence debate that AI will benefit humanity.
See information on other Sylvan debates here.