Browsed by
Month: September 2022

Robot sets 100M run world record

Robot sets 100M run world record

Cassie the robot is no Usain Bolt, but did established a Guinness World Record for the fastest 100 meters by a bipedal robot. The historic time of 24.73 seconds was achieved by starting from a standing position and returning to that position after the sprint with no falls. Though staying in your lane was apparently optional.

Cassie was invented at the Oregon State University College of Engineering and produced by Agility Robotics. This achievement was accomplished through robot learning and almost a year of simulation, condensed down to a matter of weeks.

Using Stable Diffusion for compression

Using Stable Diffusion for compression

Last week, Swiss software engineer Matthias Bühlmann discovered that the popular image synthesis model Stable Diffusion could compress existing 2D images with fewer visual artifacts than JPEG or WebP at high compression ratios, though there are some important limitations.

When Stable Diffusion analyzes and “compresses” images into weight form, they reside in what researchers call “latent space,” which is a way of saying that they exist as a sort of fuzzy potential that can be realized into images once they’re decoded. With Stable Diffusion 1.4, the weights file is roughly 4GB, but it represents knowledge about hundreds of millions of images.

While most people use Stable Diffusion with text prompts, Bühlmann cut out the text encoder and instead forced his images through Stable Diffusion’s image encoder process, which takes a low-precision 512×512 image and turns it into a higher-precision 64×64 latent space representation. At this point, the image exists at a much smaller data size than the original, but it can still be expanded (decoded) back into a 512×512 image with fairly good results.

Bühlmann’s method currently comes with significant limitations. It’s not good with faces or text, and in some cases, it can inject detail features in the decoded image that were not present in the source image. (You probably don’t want your image compressor inventing details in an image that don’t exist.) Also, decoding requires the 4GB Stable Diffusion weights file and extra decoding time that are inherent with Stable Diffusion.

Not the first time that AI has been explored as a method of compression as much as generation. Daniel Holden of Ubisoft presented an astounding paper at GDC in 2018 about using neural nets to compress animation data used in video game character animation.

Links:

Sharing things teens aren’t ready to hear

Sharing things teens aren’t ready to hear

Some adults share things they think teens aren’t ready to hear – but it is actually some solid advice for all ages. Do you have some of your own to add?

  1. Social media is not reality and your entire life should not revolve around it.
  2. Everything you’re doing now as a teenager will be utterly cringe to your children.
  3. You won’t feel different when you’re older, or have kids. You’ll just be you – it’s just weird
  4. Today’s eyebrows are tomorrow’s clown makeup
  5. In 15 years you’re going to think kids have gone too far and they’re going to think you’re old fashioned
  6. Getting good at stuff will take time. Sometimes a lot of time. And sometimes you’ll spend lots of time on somethinig and you still won’t get good at it. That’s the human experience. Some things you struggle with will come very easily for other but some things they struggle with will be very easy to you. Don’t be made someone else possesses a skill you don’t, and don’t be a jerk for possessing skills many others don’t.
  7. Nobody wants to hear whatever TikTok you’re watching. Buy some damn headphones.
  8. Being controversial isn’t the same as being interesting.
  9. School has a system for keeping you from falling behind. Life does not.
  10. Just because you f*-ed something up does NOT mean you’re a f*-up
  11. Things will likely take significantly longer to achieve than you think.
  12. “Life is not like a video game where you just keep leveling up. Sometimes the thing you built will fall apart and you will have to repeatedly do the same thing over and over. However, don’t beat yourself up about it – this is normal. With experience you will also become more adept at facing and resolving problems, so each time the same problem repeats you will be better at solving them.” – This one I sort of agree and disagree with. If you’re on a treadmill of repeated failures, especially failing the same way, then there’s probably a pattern there that needs to change.
  13. Not everyone can be an internet sensation, somebody has to drive the dump truck.
  14. One day you too will be old and uncool. It’ll happen faster than you think.
  15. Just because it’s new to you doesn’t mean it’s new.
  16. As you get older you just keep realizing how dumb you were last year.
  17. That heartache you’re going through? It consumes everything now but it will be nothing but a footnote in the future. You’ll rarely think about it later and when you do it won’t hurt you. It’s hard to hear that your pain isn’t the worst in the world when you’re feeling it, but it does help to know that it won’t mean as much as it does in this moment.
AI entry wins State Fair

AI entry wins State Fair

Théâtre D’opéra Spatial – Jason Allen

Jason Allen spent roughly 80 hours working on his entry to the Colorado State Fair’s digital arts competition. Judges awarded him first place, which came with a $300 prize.

Allen’s victory took a turn when he revealed online that he’d created his prize-winning art using Midjourney, an artificial intelligence program that can turn text descriptions into images.

The revelation (which he made even when he dropped off the submission) has sparked intense debate as to what constitutes art – to the point that it’s spawned people saying it’s the death of art. As AI enters each field, and then soon does better than even the best people (which we’ve seen time and again), we’re going to have to re-think society and work. When everyone can use AI tools to make the best of anything in the world – it’s going to have profound impact on everything.

I suspect we’ll see this in our lifetimes based on the current rate of AI growth and capabilities. I, like many others, wonder if we’ll survive it.

Read more about the controversy on the Smithsonian website.

https://twitter.com/OmniMorpho/status/1564782875072872450?s=20&t=mKjO0RXPHwpTSu3YmlyZlA
Sailing into a Hurricane

Sailing into a Hurricane

Saildrone Explorer SD 1078 is an unmanned wave-rider that dove into the heart of Hurricane Fiona in the Atlantic Ocean this week. It sent back some wild video as it battled 50-foot (15-meter) waves and winds in excess of 100 mph (160 km/h).

These little Saildrones are increasingly sailing around the world’s oceans collecting valuable data. Below is SD 1078’s eerie video from inside Hurricane Fiona.

Decentralized Autonomous Organizations

Decentralized Autonomous Organizations

A lot of people like to talk about the Great Resignation and alternative working models. Enter DAO’s and Web3. Decentralized Autonomous Organizations promise to make you your own boss and worker. They seem to appeal to folks interested in DeFi movements. So what are they and how do they work?

First, you join a DAO – usually by by purchasing a token. Then you can then start participating and contributing to the community forum (usually a discord channel) and voting (using Snapshot or similar). As one’s reputation grows, the DAO community may reward you based on their discussion and participation KPI’s. Once you prove your reputation, you might start contributing to the core DAO project. At this point, you might be able to participate in completing a bounty: a small task. After your reputation grows enough, you might even be picked up for a full or part time role. Part and full-time positions in DAO’s are still rare and people that earn money often participate in several of them at the same time. It’s similar to many other hustle culture jobs. This tweet fits the experience:

The draw is that you can quickly put together a community of like-minded people interesting in pushing forward an idea or project – and let people contribute to it and work on it as they can. You can work when you want from wherever you want.

There are some serious downsides besides the difficulty of making a living doing it. Many people are ok with low/no pay because they’re more interested in the community organization around an idea than the pay, but organization can be difficult. Coordinating work and incentives so everyone is focused towards a solution can be problematic. Further, there are no guarantees. You may never get paid. There is no retirement contributions, stock, employee gym, no medical coverage, no sick/maternity/paternity or annual leave. All the work you do is in a legal grey area of ownership since you are neither an employee nor contractor – but liability claims might apply to every participant. Harassment, discrimination, and other workplace issues can also become problematic as the DAO might exist in countless different countries and jurisdictions. Still, it hasn’t stopped them from forming.

The takeaway

I get the idea, but have serious doubts about these organizations – just like I had serious doubts about cryptocurrency promises. The reason we have laws about companies is to protect employees from unscrupulous bosses/exploitation, help resolve conflicts between owners that want to go different directions, protect a company’s hard work from competition, defend themselves from patent/IP/financial attacks, and to hold companies liable if they do harm. Their ideas and work could easily be stolen by others and it’s questionable that anyone could protect their IP. It would be interesting to see what happened if there were copyright claims against a DAO’s work if it was plagiarized. They could easily be infiltrated and overtaken by subversive elements. They could be run by competitors to destroy a competitor’s market or by foreign powers interested in subverting a government, legal system, or its financial systems to start a revolution/cause massive riots. The reality is that enabling people to overthrow existing systems can be used for evil just as quickly as for good – and we don’t seem to have nearly as many controls on the former to justify the latter.

I worry that, just like crypto-currency, we’re going to have to re-learn why these laws exist and just changing names doesn’t change human nature. Lets hope we have more skepticism and learn our lesson before another entire population of people get run through the ringers like happened with the completely predictable crypto collapse.

Other links: