The Artificial Intellectual

A lot of people worry about how computers will destroy humanity once they become sentient.  Perhaps they’ll save us.

This looks promising. It is a nonprofit doing AI research so that the results can benefit the world instead of stakeholders. Elon Musk is a co-chair.

Amazon are talking about delivery drones again. The sci-fi futurist in me finds this really cool but disturbing. There is the issue of replacing human jobs with robots (I will speak more for both sides of that in a future post). The impact to privacy really worries me, though. Once these things grow ubiquitous, what is to prevent law enforcement agencies from requiring access to their video feeds? There have been attempts to do so with webcams and mobile phones already. Many police cruisers constantly record license plates while on patrol, storing their locations and timestamps in a giant database for future reference. Jeremy Clarkson is cool, though.

Happy Bleak Friday. Enjoy some futuristic predictions from Ray Kurzweil.

I think that robots should emulate the interaction style of the objects they are designed to, well, emulate. If a robot is intended to replace a machine in a factory, it should have pretty simple conversational structures. If it is designed to replace a human (and interact with humans), then design it to communicate like a human so that humans will instinctively know how to deal with it.

Dr. Thomas G. Dietterich of Oregon State University discusses some popular concerns regarding artificial intelligence. Two interesting ones that need more attention are the socioeconomic impact of the robotization of jobs and the consolidation of tech capital.

The title is a bit misleading, since the algorithm is not actually presented, but this is an interesting read about an attempt to remove bias in a large-scale study by using algorithmic analysis instead of human analysis of sunject matter. As we heard in a discussion from Science Friday last week, however, bias can still emerge in the algorithm design.

The Marines have been testing a quadraped robot from Boston Dynamics, a company owned by Google/Alphabet. While this small (160 lb.) version is not an AI system, it is definitely a precursor. I got to see some of BD's bots demonstrated at a DARPA conference earlier this year, and they are both powerful and creepy. Though they are currently only proposed for scouting and hauling equipment, it is not a leap to imagine these things armed and sent into combat. You should be afraid.

Researchers have found a way for living plants to absorb conductive materials and pull them up their stems, creating a sort of internal wiring harness. This can then connect to sensors or other devices throughout the plant. Uses suggested include monitoring plant health, controlling hormone release, or even drawing power from photosynthetic activity. I, for one, would much rather power my supercomputer off a forest than a coal plant.

If you are into computing power (and you should be) here’s a teaser about the forthcoming Top500 supercomputers list. Spoiler: China is winning.

Science Friday jad a good segment today regarding bias in algorithms. They showed several examples of how, since bias is a trend, making decisions based on unreviewed trend analysis can sneak in a bit of bias. This is definitely an area of concern as big data grows in popularity and ease of use.

Yes, the latest iteration of the machine that beat grandmaster Kasparov can now be used to predict the toy your child has been most conditioned to covet this holiday season. Watch the video. It sounds like Watson is being employed for trend analysis, but they keep throwing around the term “cognitive” to describe the venerable supercomputer’s powers. I will have to read up on Watson sometime soon. I cannot help but feel a bit sad that this is the best use case.

There is a lot of fear concerning the future of the job market in the wake of automation. Learning new things seems to be our best hope for remaining relevant. The old models of education worked well when jobs were not being rapidly phased out and replaced with jobs impossible to foresee.

I read this book last year when all the hype came out about a female sci-fi author’s debut novel winning a Hugo, Nebula, and Arthur C. Clarke award.  The hype was deserved.  I have not yet figured out a way to tell people about why I love this book without spoiling my favorite part, but rest assured that Ann Leckie does something very creative and compelling with artificial intelligence.  This book has action, character development, plot, tech, philosophy, and everything you could ever want.  You really should read it.

I was really excited to find this page, and then really saddened when I found it incomplete.  As soon as I read about pancomputationalism the first time it struck me what a great basis for a religion it would make.  Apparently I am not the only one to have had that thought.  The site is worth checking out for the priest’s name alone.

Wow, I have been reading about pancomputationalism (that everything is a computational process) and it turns up in XKCD.  In this instance, DNA is the output of a billion years of optimization.

Here’s a fun (though negative) recap of several AI’s from 1970’s and 80’s sci-fi movies and TV shows. It does not offer much in the way of elevating the conversation, but it might help you pick out some old favorites to rewatch.