AI Talk: Unethical AI, hearables and fruit-picking drones

February 26, 2021 / By V. “Juggy” Jagannathan, PhD

This week’s AI Talk…

“Who should stop unethical AI?”

That is the intriguing title of an article in The New Yorker. There is now a spotlight on problems created by indiscriminate use of AI algorithms. The article starts with a maelstrom of protests over a project called “Speech2Face,” a program that paints a face after listening to a person speak. The argument is whether such work should even exist?” Certainly, the work and its use cases are questionable.

The fact is, for general purpose AI research, there is no such thing as an Institutional Review Board (IRB) process in academia that is mandatory for any research involving human subjects. IRB investigators would ensure the work undertaken is ethical and that no harm could come from the research. Katie Shelton, a scientist at the University of Maryland and chair of a research-ethics committee in one of the Association for Computing Machinery’s (ACMs) Special Interest Groups, calls out four potentially problematic areas:

  1. AI that can be used to surveil a population, like facial recognition and location tracking
  2. Technologies such as “Speech2Face” which may be inherently biased and cause problems
  3. Automated weapons research
  4. Alternate reality, such as fake news, voices, images or video.

These concerns can overlap in any given application, leading many major conferences to focus on creating some AI guard rails. The Association for Computational Linguistics, the Association for the Advancement of AI and NeurIPS have all instructed reviewers to incorporate ethical impacts. AI research authors are also encouraged to submit their take on the ethics of their research and its impact on society. These are steps in the right direction, but they are unlikely to staunch the flow of problematic work. Perhaps requiring an IRB for all academic AI research may be useful as it would require getting permission BEFORE one embarks on the work. Industry research? That’s another can of worms!

Hearables

We have all heard about wearables, but have you “heard” about hearables? I found this recent article in Forbes interesting – it explores the whole market segment of hearables. Essentially, everything you can do with your voice and your earbuds! Headphones and earbuds have become more or less a staple commodity in this pandemic-induced Zoom-world, but their use goes beyond noise cancellation and hands-free calling.

There are a lot of potential apps being investigated that go beyond the basics. For instance, sensing your mood and playing music to match. AI algorithms determine if you are stressed and, if you are, suggest soothing music to play. There is even an app which can figure out if you have COVID-19 based on your voice. Voice technology can also detect if you’re sick and suggest remedies. It can even recognize who you are. When voice is used as a biomarker and an app in your ear promotes health and wellness—you have a hearable!

This decade will go down as the decade when voice became the primary interface to your world of objects, from computer, to phone, to car, to toaster oven! In the process, your voice may reveal a whole lot of things to the folks listening. Privacy will continue to be an issue, but a lot of venture capital is chasing this market along with all the usual suspects—Google, Amazon, Apple and Microsoft.

 

Fruit picking drone

Flying Autonomous Robot (FAR) is a new class of drones that has emerged as a solution for fruit farmers. The acute labor shortage in the farm sector now has a robotic shot in the arm. These flying drones from Tevel use vision technology to identify and pick fruit right off the tree! Of course, this also means the migrant workers who relied on this work as a source of income may have to look elsewhere.

I am always looking for feedback and if you would like me to cover a story, please let me know. “See something, say something!” Leave me a comment below or ask a question on my blogger profile page.

V. “Juggy” Jagannathan, PhD, is Director of Research for 3M M*Modal and is an AI Evangelist with four decades of experience in AI and Computer Science research.


Listen to Juggy Jagannathan talk AI on the ACDIS podcast.