Skip to content
SF AppWorks Nov 23, 2018 2:39:00 AM 4 min read

The Wonder - AI Bias, Solid-State Propulsion | SF AppWorks

A Weekly Snapshot of Life-Changing Technology

Happy Friday!

Happy Thanksgiving. This week, in California, we’re thankful for rain that is pushing away hazardous smoke and putting out the last of the vicious fires that claimed 87 lives – a toll that is likely to go up as rescuers sift through the wreckage. It was the deadliest and most destructive fire in California’s history, and there was little warning.


But why is it so hard to issue a fire warning?

We have tornado warnings, flood warnings, and hurricane warnings, but in the last few major fires to strike California, very few people were notified that a growing wildfire was heading their way.


Axios science Editor Andrew Freedman addressed the question in his weekly science newsletter in three parts:


  1. Humans are a wildfire threat multiplier. Though no single factor – not climate change, forest management, or building practices – is responsible for the deadly blasts the state is now seeing, their combination is making an already dicey situation far worse…Longer-term climate change and population growth are combining to increase wildfire risk in California and the American West. 
  2. There is something wrong in how we’re handling fire emergencies. Forecasters knew days in advance that conditions would be extremely conducive to wildfires, but unlike severe thunderstorms and tornadoes, the journey from forecast to a warning being delivered is convoluted. Though there is a ‘fire warning’ system, warnings are seldom issued in part because of fear of the panic that it will cause. 
  3. We’re on the cusp of better wildfire forecasting. Unlike tornadoes and hurricanes, there have been no major campaigns to study and understand wildfires. Perhaps this is because they’ve tended to burn away from populated areas, until now. There is also no tried and tested computer modeling suite for fire forecasts, but they are rapidly improving. Weather forecasters are successfully anticipating days that have the potential for disaster if a fire were to ignite.


Bottom Line – it’s not a technology issue, but one of awareness. New safety systems tend to arise only after disasters. 


Whether you agree or don’t agree, hit reply and tell me why.



Three Neat Things
  1. Walk Off. Google is working on VR shoes that let you walk forever in your living room.
  2. Call Screen. A new feature for the Pixel utilizes Google Assistant to pester spam callers with automated questions while you watch with satisfaction.
  3. Heads Up. Leaked pictures of a columnar transparent OLED screen from Samsung suggest the company is working on a holographic display.

Now on to the Wonder.


AI Bias


vectorial image of stats


MIT Computer Science AI Lab (CSAIL) researchers have created a method to reduce bias in AI without reducing the accuracy of predictive results.


Read More>>

Andrew’s Take: We’re a long way from eliminating human bias and now we have to work on computer bias too? Better get cracking. The good news is – studies like this show an effort to get ahead of the problem, rather than wait for it to become systemic.


Darius’s Take: AI Bias might not be a big deal today, but as AI gets intertwined with healthcare, criminal justice, and automation at scale, it could negatively impact society as a whole.


Solid-State Propulsion


A team of researchers at MIT have managed to build the first aircraft powered by an ionic wind – a propulsion system that requires no moving parts.


Read More>>

Andrew’s Take: It’s exciting and a milestone nonetheless, but we’re far away from seeing silent drones and platforms that can suspend in the air indefinitely. The science behind it is sound, but we’re not technologically ready to create enough energy to provide any reasonable amount of thrust.  


Darius’s Take: True, the first iteration has a pretty low efficiency rate, but it’s a technology with lots of potential that can already be applied to existing products while they progress with the next iteration. A good example would be to use the system to steer and to extend the range that gliders have, which would offer a good development and testing environment to build a system that might eventually be able to propel itself.


A Freaky Child Android Head


Osaka University Researchers created a robotic child head called Affeto that can recreate human-like facial expressions.


Read More>>

Andrew’s Take: I’m terrified by the implications of this. I watched the video and, though I fully understood I was watching a robot, I felt the kind of empathy that a cute kid trying to make faces at me might elicit. Wherever the future takes us, we’re going to start having emotional attachments to objects and I have no idea how, as a society, we’re going to deal with that.


Darius’s Take: I never thought that emulating facial expressions was that much of a technical challenge, but quite a bit has to come together. We tend to associate robotic faces with, well, robots, and I always assumed this was by design. Thinking more about it, the human face has 43 muscles which lead to quite a number of possible combinations to get the right facial expression and, more importantly, to get it to look natural.


Colorado State-ment...


Colorado has adopted California emissions standards, joining California and other states in moving pre-emptively to avoid any wearing of federal emission standards.


Thanks for reading! We’ll see you next week.


-Andrew and Darius 


Did you enjoy The Wonder?