Speech synthesis, based on artificial intelligence, makes it possible to generate spoken text from a written text.
“The border between true and false is becoming more and more blurred, and whether we like it or not, these intelligent machines are everywhere”: host Patrice Roy received on his set Thursday some of the most brilliant brains of the field of artificial intelligence (AI) in order to decode this technological revolution taking place in the world.
AI is a system that wants to learn and wants to do something that has been asked of it. To do this, she needs a goal, tools – often data – and she needs feedback. This is how AI is defined, according to Valérie Pisano, president and CEO of MILA, the Quebec Institute of Artificial Intelligence.
This technology was already tormenting scientists during the Second World War, starting with the British Alan Turing, who worked on a machine to decode Nazi messages. However, it was not until 1956, on the sidelines of a historic conference at Dartmouth, that the expression AI appeared.
Today , if AI is on everyone's lips, it is mainly because of OpenAI's ChatGPT conversational robot.
The level of intelligence of ChatGPT, researchers thought it would take decades.
A quote from Yoshua Bengio, founder and scientific director of MILA
Open in full screen mode
From left to right: Valérie Pisano, President and CEO of MILA, and Joëlle Pineau, associate professor at the School of Computer Science at McGill University and vice-president, AI research, at Meta.
< p class="StyledBodyHtmlParagraph-sc-48221190-4 hnvfyV">Joëlle Pineau, associate professor at the School of Computer Science at McGill University and vice-president, AI research at Meta, still wants to be reassuring. ChatGPT is still far from having ingested all human knowledge […] The model is focused on understanding language, so it does not understand music, images, and other information that we are capable of understanding, says- t-elle.
As Valérie Pisano reminds us, what is special with ChatGPT is the feeling of being in a relationship. When she questions the chatbot, she feels the need to be polite, thanking it and congratulating it for its good deeds.
LoadingQuebec will use gag order to pass Bill 15
Quebec will use a gag order to pass Bill 15
ELSE ON NEWS: Quebec will use a gag order to pass Bill 15
This is precisely what Turing warned was the next big step, when we were going to interact with machines without being certain of knowing whether acts more like a human being. That would mean that we have entered a new era.
A quote from Valérie Pisano, President and CEO of MILA
These systems are developing an understanding of the world that is not yet as good than ours, but which is almost encyclopedic, underlines Yoshua Bengio.
In this two-hour special broadcast in front of an audience, Patrice Roy receives on his set specialists and world luminaries in the field of artificial intelligence to answer our questions, our concerns, our hopes.
A spectator, Lucie Turmel, wonders if AI could be used during health consultations to compensate for the difficulty of accessing a doctor.
We are not there today, one of the reasons is that systems can hallucinate. And in the case of medicine, it can be serious, answers Yoshua Bengio.
However, he adds that AI could be used to understand the very foundations of biology, which could help scientists develop drugs, among other things.
It could be a revolution for medicine in the years and decades to come.
A quote from Yoshua Bengio, founder and scientific director from MILA
Open in full screen mode
Pioneer of deep learning, Yoshua Bengio is the founder and scientific director of MILA.
And new drugs or new molecules could be developed quickly – a cycle of 6 months to 1 year instead of 25 years – thanks to the significant computing capabilities of AI, according to Joëlle Pineau. She adds that tests could also be carried out synthetically rather than in the laboratory.
Artificial intelligence has the potential to replace, to a certain extent, creative professions, including those in music. Could she even become the next Mozart? Maybe not, replies Joëlle Pineau, but something else, yes, which could speak to another generation.
Generative AI generates, and it draws inspiration from everything it knows about us, our music, but also our poems and our blogs. And it will become more and more sophisticated.
A quote from Valérie Pisano, president and CEO of MILA
But you still need to know who to grant the copyright to a musical track generated by an AI. Several artists' unions in Quebec, including the UDA, consider it inconceivable that content generated by AI could benefit from the status of works protected by law. In particular, they require that it specify that an author must be a human being.
Scientific journals have notably positioned themselves in relation to this questioning, according to Joëlle Pineau, wanting the use of AI to be included in the acknowledgments, and not among the authors of an article.
< p class="StyledBodyHtmlParagraph-sc-48221190-4 hnvfyV">After all, artificial intelligence responds to commands from human beings, adds Bruno Guglielminetti, digital media specialist.
Open in full screen mode
Bruno Guglielminetti is a specialist in the digital world.
These ethical questions are just the tip of the iceberg of the dangers linked to advances in artificial intelligence.
Misinformation is also part of it. Systems as we know them today can be used to easily generate fake news that will be easily convincing, and then propelled by recommendation algorithms on social networks. It can be damaging for the quality of democratic information, explains Jocelyn Maclure, ethicist and professor at McGill University and president of the Commission on Ethics in Science and Technology.
Democracy is already experiencing upheavals since the election of Joe Biden in the United States in 2022, contested by many supporters of Donald Trump. And the year 2024 will be important for democracies around the world, since more than 4 billion people will be called upon to vote, according to Valérie Pisano.
Knowing how to discern truth from falsehood is essential in the democratic process.
A quote from Valérie Pisano, president and CEO of MILA
Among the possible solutions, Jocelyn Maclure notes the establishment of mechanisms by the different platforms to clearly designate whether content has been generated by an AI, as well as obligations to delete false or misleading content.
< source srcset="https://images.radio-canada.ca/q_auto,w_700/v1/ici-info/16x9/jocelyn-maclure-intelligence-artificielle-ia-emission-speciale.jpeg" media="(min- width: 0px) and (max-width: 1023px)">Open in full screen mode
Jocelyn Maclure is an ethicist, professor at McGill University and president of the Commission on Ethics in Science and Technology.
The ball is in the governments’ court. What scares me the most are the systems that will exist in a year, in two years or three years. And regulatory safeguards must be put in place so that the government has some control. These systems will be even more efficient, which can be even more useful, but also more dangerous in the wrong hands, insists Yoshua Bengio, who pleads for counter-powers.
François-Philippe Champagne, Canada's Minister of Innovation, Science and Industry, considers that Canada is a step ahead in regulating AI. We were the first country in the world to have a national strategy on AI, one of the first countries in the world to have tabled a bill to regulate AI, and one of the first countries in the world, along with the United States, to have a code of conduct voluntary, he says.
If you apply for a job, a loan from the bank or a car insurance policy, you don't know today if it was an algorithm that decided or a human. This is worrying.
A quote from François-Philippe Champagne, Canada's Minister of Innovation, Science and Industry
The obligation of transparency would be the least of things in a possible regulatory framework, specifies the minister.
What is certain is that artificial intelligence must serve human beings, and not the other way around, says François-Philippe Champagne. An opinion shared by Yann Le Cun, vice-president and scientific director of AI at Meta, whose vision of this technology is more optimistic. We have this sort of idea that, necessarily, if we have a system or an intelligent entity, it wants to take control. But it's a projection of human nature that has no reason to make, he says.
I have confidence in the ability to society to do what is necessary to bring out the beneficial effects, he adds.
