AI talk: Applications galore

June 17, 2022 / By V. “Juggy” Jagannathan, PhD

This week I sample a collection of new AI-based solutions that hit my radar.

Interview assist

I was fascinated by a new tool from Google, which takes conversational assistant technology to another level. What is the new tool? They call it the “Interview Warmup.” How does it work? Well, Google has curated a set of questions from industry experts in a range of jobs, mostly related to IT and tech. A job candidate can then practice answers to the curated questions presented using a voice interface. The answers are then transcribed into text. A natural language processing application provides insights on better word and phrase choices to improve the candidate’s answer. The candidate can try different ways to answer and the app essentially provides a safe space to practice. The app is in beta release. Interesting!

Dental hygiene

Electric toothbrushes have been around for a while now. We even saw AI introduced into these devices a few years back. Now, a new iteration of AI-enabled toothbrush that grades your brushing skills has been released. A pressure sensor provides feedback on the amount of pressure applied to gums during brushing. 3D, real time imaging shows which teeth you have brushed and which you have not. And, if you are not satisfied with just cleaning teeth, it will track if you have cleaned your tongue. What will they think of next?

Scarecrow reimagined

An age-old problem, now with a novel solution. The problem? How to scare away birds that feed on your crops. The solution? Drones to the rescue. Washington State University researchers have developed a prototype drone system that utilizes a machine vision and auditory system to literally scare away birds. You can read a complete description of this system in their published paper. Basically, the drones are configured to patrol a field of interest. When the vision system notices a bird in the field (essentially the machine vision system recognizes a bird in the drone camera video stream), the drone flies toward the bird and scares it away using a loud sound. It can also fly in various random ways to scare away the bird. Current testing methods only apply to a single bird. Dealing with a flock of birds is future work. The last sentence in their paper concludes: “In addition, the efficiency of the proposed system (with desired improvements) in deterring birds should be assessed in the field environment.” So, you won’t see drones scaring away birds for a while, but an interesting application, nonetheless.

New Apple releases

The annual Apple Worldwide Developer Conference was held recently. True to form, Apple has released a plethora of features with iOs upgrades across its devices: iPhone, iPad, Apple Watch and MacOs. You can listen to the two-hour preview program here. What caught my attention is the updates to the health category. Apple now has an upgraded sleep app that will categorize your sleep stages as REM sleep or deep sleep, etc. That kind of tracking helps determine sleep quality and can help diagnose sleep disorders. Another upgrade is to atrial fibrillation (AFib) detection. Devices can now track all occurrences over time to give a better idea of what is happening long term. This AFib history is undergoing FDA approval. Finally, there is now a medication reminder app along with alerts related to drug interactions. If you want a detailed, bulletized list of what Apple unveiled, check out this excellent list.

Sentient chatbot

The Washington Post carried an interesting story about a Google engineer entrusted to test its chatbot. The Google chatbot, built using LaMDA, or Language Model for Dialog Applications, is a transformer-based application platform for creating chatbots. You can read the technical paper on this model published earlier this year here. The Google engineer was convinced the program was sentient. Anyone who reads and understands the technical underpinning of this program will immediately know the claim is baloney. However, it points to the fact that the program is good enough to convince at least one individual. In that sense, the program has passed the famous Turing Test. The technical paper, in the concluding section, has this to say:

“A path towards high quality, engaging conversation with artificial systems that may eventually be indistinguishable, in some aspects, from conversation with a human is now quite likely.”

The authors also point toward the multiple risks involved in deploying such technology – one reason they have people testing it for bias, toxicity, cultural sensitivities and other aspects. But one thing this program is not is sentient. Impressively human-like is a better term.

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

“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.