Ok. It just happened. One autumn day, an artificial intelligence passed the Mensa test with a score far above the human average. It was done by o1, the model recently presented by OpenAI. And with its IQ of 120, it didn't just surpass the human average: it opened a Pandora's box of ethical, philosophical and practical questions. Do we or don't we want to decide to redefine the parameters that define so-called "intelligence" in a world where machines can outclass our evaluation systems?
The Rise of o1: The 'Bad Boy' in the AI Sanctuary
Il Strawberry Project di OpenAI has produced what the company describes as a breakthrough ingeneral artificial intelligence (AGI). The o1 model is not a simple algorithm, but a “thinking” AI that uses reinforcement learning and thought chain processing to mimic human reasoning and facilitate problem solving through deductive inference.
The test results? Impressive to say the least:
- 89th percentile in doctoral level applications,
surpassing most humans; - 83% of problems solved in the qualifying exam for the International Mathematical Olympiad;
- IQ of 120 in a Mensa test.
“Talents” that place o1 well above the human average, which stands between 85 and 115 IQ points. I repeat, as I said the day after the passing the Turing test from GPT-4: what to do?
Beyond the Mensa Test: The Implications of a “Brilliant” AI
o1's success in the Mensa test is not just a technological achievement, as mentioned, but a turning point that raises profound questions about the future of humanity and technology. Yuval Noah Harari, historian and philosopher, in his book “Nexus: A Brief History of Information Networks from the Stone Age to AI” describes AI as an agent, fundamentally different from any other technology invented by man.
“AI is the most dangerous technology we have ever created,” says Harari, emphasizing its unique ability to operate autonomously and evolve.
Potential and dangers: the double face of AGI that wins the Mensa tests
The potential of the AGI to come (and they say which will come soon) is vast and promising. It could:
- Personalize education;
- Reinventing industries for sustainability;
- Accelerate scientific research;
- Address food and water shortages on a global scale;
- Mediating human conflicts (I don't believe in that much here).
The other side of the coin? Terrifying, needless to say. If implemented improperly, this technology could also:
- Producing fake news, reconstructions and even more misleading information;
- Promote totalitarian and repressive regimes;
- Increase the destruction of job roles beyond measure;
In other words, a future that no longer includes us.
The Race for AGI: A Competition Without Rules?
I assure you: despite my cautiously optimistic approach, the speed with which Big Tech companies are pushing AI into the mainstream raises concerns even for me. Aside from a few clear warnings, There is a lack of adequate guardrails and governments, too those willing, struggle to keep up with the speed of change.
What would it take to truly have an AGI at the service of humanity?
- The exact opposite of what I perceive now: a gradual implementation in controlled environments;
- A clear alignment of AGI systems with human-determined goals;
- A common agreement between governments on development and implementation parameters;
- An ongoing debate on the direction of AGI research, with strong ethical and sociological.
The future is already here: are we ready?
OpenAI’s o1 model and its exploit in the Mensa test represent an important milestone, but as Harari points out, there is an inherent danger when a human invention becomes an autonomous agent, capable of ignoring its creators in its pursuit of knowledge and growth.
The challenge that awaits us is not only technological, but deeply ethical and philosophical. As a society, we must face crucial questions. I ask them to myself and to you:
- How can we ensure that AGI remains aligned with human values?
- What role will humanity have in a world where machines can surpass us in “intelligence”?
- How can we harness the potential of AGI while minimizing the risks?
The Mensa test passed by o1 is just another sign of a new era. The real test will be how we respond to this technological revolution. Our response will show, beyond any test, what true intelligence we have.
