H+ Weekly - Issue #89
We must become cyborgs, says Elon Musk. AI learns to be aggressive while another AI learns from less data than deep learning AIs. Patent office hands win in CRISPR battle to Broad Institute. And as always, other news and links about AI, robotics, and biotech!
If you enjoy the newsletter, please consider sharing it on Twitter, Facebook, etc! Really appreciate the support!
MORE THAN A HUMAN
Elon Musk is convinced we need to become machines or risk being replaced by them
Elon Musk repeats his view that the only way for humans to stay relevant in the age of AI is not to fight them, but to join them by merging ourselves with machines.
DARPA: We’re Moving to Merge Humans and Machines
Elon Musk: "We need to merge with machines!". DARPA: "We are working on it!".
The ethical dilemma of designer babies | Paul Knoepfler
Creating genetically modified children is no longer a science fiction fantasy; it's a likely future scenario. Biologist Paul Knoepfler estimates that within fifteen years, scientists could use the gene editing technology CRISPR to make certain "upgrades" to human embryos -- from altering physical appearances to eliminating the risk of auto-immune diseases. In this thought-provoking talk, Knoepfler readies us for the coming designer baby revolution and its very personal, and unforeseeable, consequences.
Sony has patented a contact lens that is blink powered and records video
Apparently, Sony has patented a contact lens that can record video and you control it just by blinking.
Google's New AI Has Learned to Become "Highly Aggressive" in Stressful Situations
Guys from DeepMind were curious how AI will react in a competitive environment with scarce resources. They have created a simple game where two agents had to collect apples. Everything was going ok until the number of apples dropped down. When it happened, the agents started to be aggressive towards each other, proving that AI can become hostile if the conditions are right.
AI Software Juggles Probabilities to Learn from Less Data
Deep learning allowed machines to do things we though are decades away or even beyond computer's reach. As great as it is, deep learning has a flaw - it requires a vast amount of data to learn. Gamalon, a Boston-based startup, uses a different approach called Bayesian program synthesis to build algorithms capable of learning from fewer examples.
Tony Stark Has Jarvis. And Now IBM Has Havyn
This is a cool story. Evan Spisak built in his spare time with his 11-years old son Havyn - a voice assistant for cybersecurity. It started as a simple project on Raspberry Pi backed by IBM's Watson and it's growing now to become something much bigger.
Will Superhuman Intelligence Be Our Friend or Foe? | Ben Goertzel
As Ben Goertzel says, you cannot rule out the possibility that superhuman AI will be a threat to humanity. But he says there is a way to minimalize the AIpocalypse by making sure general artificial intelligence belongs to everyone, is open source and it's used to the benefit of whole humanity and not only for the benefit of one company or nation.
Neural Network Learns The Physics of Fluids and Smoke
AI researchers from Zurich decided to teach a neural network how to generate realistically looking smoke and water effects by showing it a lot of videos how smoke and water behave in a real life. It might sound like a pointless research, but the team was able to generate convincing looking effects in a shorter time than with using traditional approach and their work might find use in computer graphics and games.
Is AI a Threat to Christianity?
Here's an article that might make you ask different types of questions regarding AI. For example, if we create superintelligent machines and accept them as conscious beings, could they be baptized? How would the creation of such machines impact our current religious systems? Would it pray (if you read Issac Asimov's I, robot you know it is possible)?
How to Keep Your AI From Turning Into a Racist Monster
Just make sure your training data is not biased in any way. But it's easier to say than done.
How Robots Helped Create 100,000 Jobs at Amazon
"Robots are taking our jobs!", you can read or hear when people talk about automation. It's true that robots will take some jobs, but at the same time, they will make space for new ones. Here's an example from Amazon. By using more robots in its fulfilment centres, Amazon was able to drive the prices down, which lead to increase in sales and demand, which resulted in hiring more people.
Robo-Bees Could Aid Insects with Pollination Duties
Mini drones sporting horsehair coated in a sticky gel could one day take the pressure off beleaguered bee populations by transporting pollen from plant to plant, researchers said.
Tying Knots With Industrial Robots
It's so cool to watch! I like when the robot was showing the knot. It looked like wanted to say "hey, human, look what have I done!".
Engineers Have Created Biocompatible Microrobots That Can be Implanted Into the Human Body
Soon, there might a biocompatible robot made from a hydrogel without any toxic or big components roaming in your body delivering drugs.
Patent Office Hands Win in CRISPR Battle to Broad Institute
The dispute between researchers at UC Berkeley and the Broad Institute over the invention of the powerful gene-editing technique has been decided. This ruling will have major consequences for all companies and startups that jumped into CRISPR and their choice of whom to take the CRISPR license from.
How to build your own bio-bot
A year ago or so, researchers showed a bio-bot - a tiny robot powered by real muscles. Now, these guys open sourced their research, so you can build your own bio-bot if you want! I already added it to my list of projects to do.
U.S. science advisory committee supports genetic modification of human embryos
This week, a panel of experts from the National Academy of Science released a report endorsing the research of using CRISPR to edit human genome, but there are some caveats. The experts endorse editing genes in somatic cells (the cells that aren't passed to the next generation), but they heavily cautioned against the use or CRISPR on germline cells—ones that can be passed down to future generations—as well as on human embryos. Research in this area was still in its infancy, they said, and more work needed to be done to better understand the risks before any experiments could proceed.
Follow H+ Weekly!