This week – a closer look at AlphaZero; how to write “programs” in bacterias; the gene patent question; and more!
More Than A Human
Here’s a video of Pollyanna dancing. She’s missing her right foot which was replaced with one of those blades prosthetic used by Paralympic athletes.
Researchers have demonstrated that lab-grown artificial ovaries could provide a better way of providing hormone replacement therapy for women. In a study published in Nature Communications, engineered ovaries were found to perform better than hormone therapy drugs in terms of improving bone and uterine health, as well as body composition.
DeepMind looks back at 2017 and reviews the year. “We are proud of all of our progress in 2017, but know there is still a long way to go”, writes Demis Hassabis and his team. DeepMind produces some outstanding results in 2017, culminating with the release of AlphaZero, which mastered the game of chess in just 4 hours without knowing anything about the game. DeepMind also started to focus on ethical aspects of AI and the technology they help to develop.
Here’s a more critical view on AlphaZero. Jose Camacho Collados points out flaws in DeepMind’s paper describing AlphaZero, like availability and reproducibility of the results. The main point of this article can be summed in this quote: “It is actually responsibility of researchers in this area to accurately describe and advertise our achievements, and try not to contribute to the growing (often self-interested) misinformation and mystification of the field”.
Some people are saying that the first alien intelligence we encounter will not come from outer space but from our computers as an AI. When you look what choices an AI makes, you might understand this point. Demis Hassabis, the CEO and founder of DeepMind, said this about AlphaZero at NIPS conference: “It doesn’t play like a human, and it doesn’t play like a program. It plays in a third, almost alien, way.”
“The classic Deep Learning training procedure is one of the crudest teaching methods that one can possibly imagine” , writes Carlos E. Perez. In this article, Perez brings closer an idea of Embodied Learning. Embodied Learning argues that intelligence emerges from interacting with the physical environment. The proponents of this approach to AI argue that there is no better way to learn a new subject than to allow the AI a way to interact with the subject and to discover its responses.
In this blog post, engineers from Google present Tacotron 2 – an AI system that generates human-like speech from text. It also has a cool name.
Another game where AI become better than humans.
When people discuss AI ethical question, many of them look at the problem from the Western philosophy point of view. But that’s not the only philosophy we have here on Earth. Other philosophies, like Buddism, can shed a new light on AI ethical concerns and help us better understand AI’s role in the world and how it should interact with humans.
Researchers from Cornell University experiment with putting neuromorphic chips into a tiny, insect-like robots to give them some intelligence. Because these robots are tiny (only 3 centimetre wingspan and weigh only 80 milligrams), they cannot use traditional computers. Here’s where they see neuromorphic chip coming in. Those chips mimic how neurons work. They are smaller and more efficient than normal chips and one day, the researchers hope, they will be the brain of small, insect-like robots.
If you live in San Francisco’s Mission district, you can get your food delivered by a robot named Marble.
Keller Rinaudo wants everyone on earth to have access to basic health care, no matter how hard it is to reach them. With his start-up Zipline, he has created the world’s first drone delivery system to operate at national scale, transporting blood and plasma to remote clinics in East Africa with a fleet of electric autonomous aircraft.
Genetic engineering is already full of ethical questions. Adding patents into the mix makes everything even more complicated. You’d probably don’t mind to see a patent for a genetically modified apple or rice but what about a patent for human genes? Or a patent for an entire human being? Wendover Productions (a very good Youtube channel btw) explores this problem in their latest video.
“It’s not the big explosion that’s going to transform your world overnight,” Nathaniel Comfort, a historian of science, said. Instead, “it’s the incremental increases of knowledge and in technology that’s going to really change the way we lead our lives.”
Gizmodo lists the most noteworthy breakthroughs in genetics in 2017. The list includes FDA approving first “living” therapies, restoring a seven-year-old boy’s skin, editing DNA within the human body, editing human embryo and more.
Did you ever wonder how to write a “program” for a living bacteria? In this video, Christopher Voigt explains in details how he and his team developed a library of gates that can be interconnected to build genetic circuits composed of a series of said gates that respond to a specific input with a specific output.
In this article, Tessa Alexanian shares the biosecurity lesson she learned while participating in iGEM, an international competition for students dedicated to the advancement of synthetic biology.
The lab-grown meat is coming. It’s good for the planet, and surveys show that significant numbers of people would be willing to give it a try.