Thanks to Angelo Andrew, Karen Cutter (who is leading the report release), Jennifer Lang, Richard Lyon, Zhan Wang, and Mengyi Xu. Of the 20,200 excess deaths in 2022, we estimate that: 10,300 deaths (51%) were from COVID-19 2,900 deaths (15%) were COVID-19 related, meaning that COVID-19 contributed to the death and 7,000 deaths (34%) had no mention of COVID-19 on the death certificate.įor more details please see the full article below. Excess mortality is widely regarded as the best measure of the overall impact of a pandemic since it includes deaths both directly and indirectly due to the disease. We estimate that there were just over 20,000 (12%) more deaths in Australia in 2022 than we would have expected if there had been no pandemic. Our Working Group has provided an overall summary of excess mortality in 2022. As far as I am concerned, I believe we still have a lot more work to do for AI to become truly intelligent. As Timnit Gebru would say, we're building the perfect stochastic parrots.Ĭall me an idealist, but I think AI researchers have a lot to learn from physicists and scientists in general. Forget we have no way to identify biases. Forget if we don't understand how, or even why they work. On the contrary, when the model doesn't deliver, ML scientists like to make it more complicated. When the model isn't good enough, physicists work on simplifying it in fact, their goal is to work towards a Grand Unified Theory physicists see elegance is simplicity (remember that it is Einstein who said "if you can't explain it simply, you don't understand it well enough"). In the meantime, LLMs require billions of parameters to "model" simple human language (think that English has less than 200,000 words, and than 3,000 are enough for 95% of our communications!).įurthermore, the parameters in the Standard Model are interpretable (mass of a particular particle, for example), so when you fit the model you actually learn a lot about particles. GPT-4 has one trillion.Ī model designed to explain how all particles interact, and subsequently, how the entire universe functions, requires less than 20 parameters. The Standard Model of Particle Physics has 19 free parameters. Pascal Finette Simon Sinek Adam Grant Brené Brown Uri Levine (This post was written by me, not by ChatGPT!) If you agree with this perspective, give a □ or not! Instead, we should be concerned with the uses that we humans make, or will make, with the unlimited functionalities that AI presents us, because the first and last decisions will always be ours. In this context, our concern should not be with AI. It has never been the machines! The fundamental deciding factor for the next evolutionary step is the interference of HI, through the decision of one or more humans. Instrumentalized by its own abilities, human intelligence has always been the enabler of the next disruption. HI has been present in the genesis of all revolutionary movements. Throughout history, all scientific, technological, or industrial revolutions have never been without human intelligence (HI). And what about the discussion on the risks of AI? Is it really that serious? Is the fear of an unstoppable rise of machines anchored in the belief that AI will surpass human intelligence? Are robots already developing the ability to reproduce themselves autonomously, as if AI were evolving exponentially towards the creation of an autopoietic system? It's possible, but I don't bet on it for an epistemological reason.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |