Toggle light / dark theme

Meta-Learning Machines in a Single Lifelong Trial

The most widely used machine learning algorithms were designed by humans and thus are hindered by our cognitive biases and limitations. Can we also construct meta-learning algorithms that can learn better learning algorithms so that our self-improving AIs have no limits other than those inherited from computability and physics? This question has been a main driver of my research since I wrote a thesis on it in 1987. In the past decade, it has become a driver of many other people’s research as well. Here I summarize our work starting in 1994 on meta-reinforcement learning with self-modifying policies in a single lifelong trial, and — since 2003 — mathematically optimal meta-learning through the self-referential Gödel Machine. This talk was previously presented at meta-learning workshops at ICML 2020 and NeurIPS 2021. Many additional publications on meta-learning can be found at https://people.idsia.ch/~juergen/metalearning.html.

Jürgen Schmidhuber.
Director, AI Initiative, KAUST
Scientific Director of the Swiss AI Lab IDSIA
Co-Founder & Chief Scientist, NNAISENSE
http://www.idsia.ch/~juergen/blog.html.

How humans will reach beyond biological bodies

This video discusses what it would mean to live forever or to have a greatly expanded lifespan. Would we inevitably grow bored with existence? Why do we even want to live longer?

We also discuss the philosophy of the Ship of Theseus as it applies to mind upload or body transfer. If you transfer your mind to a new body, are you still the same person? We conclude that the answer is yes if and only if you experience no discontinuities in consciousness.

Finally, we discuss a few different ways that new bodies might be formed as technology advances. Digital avatars and robotic bodies are just a few of the possibilities.

Can You Upload Your Mind & Live Forever?

Digital immortality: would you upload your mind to a computer?
https://www.dazeddigital.com/life-culture/article/60341/1/mi…-elon-musk.

Uploading your consciousness will never work, a neuroscientist explains.

AIs accurately predicted path of Hurricane Lee a week out

To make its weather predictions, it analyzes 60 million daily observations from satellite, aircraft, and ground-based reports, using what we know about atmospheric physics to determine what the weather is likely to be like across the globe over the next 15 days.

This can literally save lives — if people know in advance that hurricanes or winter storms are heading their way, they can take action to prepare — but because the model is so complex, it must be run on a supercomputer over the course of several hours, which also makes it expensive.

The AIs: AI-based weather forecasting models are starting to catch up with traditional ones, like the European Model.

Brain-computer interfaces could let soldiers control weapons with their thoughts

Imagine that a soldier has a tiny computer device injected into their bloodstream that can be guided with a magnet to specific regions of their brain. With training, the soldier could then control weapon systems thousands of miles away using their thoughts alone. Embedding a similar type of computer in a soldier’s brain could suppress their fear and anxiety, allowing them to carry out combat missions more efficiently. Going one step further, a device equipped with an artificial intelligence system could directly control a soldier’s behavior by predicting what options they would choose in their current situation.

While these examples may sound like science fiction, the science to develop neurotechnologies like these is already in development. Brain-computer interfaces, or BCI, are technologies that decode and transmit brain signals to an external device to carry out a desired action. Basically, a user would only need to think about what they want to do, and a computer would do it for them.

BCIs are currently being tested in people with severe neuromuscular disorders to help them recover everyday functions like communication and mobility. For example, patients can turn on a light switch by visualizing the action and having a BCI decode their brain signals and transmit it to the switch. Likewise, patients can focus on specific letters, words or phrases on a computer screen that a BCI can move a cursor to select.

Aspire #02 — Rohit Singla: Using AI to Develop New Treatments in Kidney Disease

Rohit Singla, an MD/PhD student, shares how his training in both medicine and engineering is allowing him to identify complex problems, understand the nuances within them and tackle those complex problems with elegant solutions that are the right fit for patients with kidney disease. Using data from over 10,000 cases, he is creating artificial intelligence tools to automatically detect microscopic changes in the kidney structure and develop new treatments to improve people’s lives.

Produced by UBC faculty of medicine development and alumni engagement.

© 2010–2021 UBC Faculty of Medicine. All rights reserved.

Saturday Citations: Hippo maxillofacial issues; implicit biases in the game of kings; AI masters Street Fighter

They announced the Nobel prizes this week! But did any of the recipients teach an AI to play Street Fighter? Here are a few of this week’s stories not yet lauded by international committees of scientists, but which we thought were pretty good:

Even if you think a galaxy is old enough to drink, you should probably go ahead and ask for ID before you serve them. The earliest galaxies in the universe captured by the James Webb Space Telescope appeared too bright, massive and way too old to have formed that soon after the Big Bang, presenting a problem for astronomers and their favorite model, the standard model of cosmology.

Recently, a team of physicists at Northwestern University used computer simulations to model galaxy formation after the Big Bang and demonstrate that (at least in the model universe) stars formed in bursts, producing light of enormously greater intensity than a modern galaxy like, say, Andromeda, where is steady and the number of stars gradually increases over time.