The Simple Fix: Door Dash Some Diversity in AI
July 25, 2024
This essay is the work of a dumb dinobaby. No smart software required.
I read “There’s a Simple Answer to the AI Bias Conundrum: More Diversity.” I have read some amazing online write ups in the last few days, but this essay really stopped me in my tracks. Let’s begin with an anecdote from 1973. A half century ago I worked at a nuclear consulting firm which became part of Halliburton Industries. You know Halliburton. Dick Cheney. Ringing a bell?
One of my first tasks for the senior vice president who hired me was to assist the firm in identifying minorities in American universities about to graduate with a PhD in nuclear engineering. I am a Type A, and I set about making telephone calls, doing site visits, and working with our special librarian Dominique Doré, who had had a similar job when she worked in France for a nuclear outfit in that country. I chugged along and identified two possibles. Each was at the US Naval Academy at different stages of their academic career. The individuals would not be available for a commercial job until each had completed military service. So I failed, right?
Not even the clever wolf can put on a simple costume and become something he is not. Is this a trope for a diversity issue? Thanks, OpenAI. Good enough because I got tired of being told, “Inappropriate prompt.”
Nope. The project was designed to train me to identify high-value individuals in PhD programs. I learned three things:
- Nuclear engineers with PhDs in the early 1970s comprised a small percentage of those with the pre-requisites to become nuclear engineers. (I won’t go into details, but you can think in terms of mathematics, physics, and something like biology because radiation can ruin one’s life in a jiffy.)
- The US Navy, the University of California-Berkeley, and a handful of other universities with PhD programs in nuclear engineering were scouting promising high school students in order to convince them to enter the universities’ or the US government’s programs.
- The demand for nuclear engineers (forget female, minority, or non-US citizen engineers) was high. The competition was intense. My now-deceased friend Dr. Jim Terwilliger from Virginia Tech told me that he received job offers every week, including one from an outfit in the then Soviet Union. The demand was worldwide, yet the pool of qualified individuals graduating with a PhD seemed to be limited to six to 10 in the US, 20 in France, and a dozen in what was then called “the Far East.”
Everyone wanted the PhDs in nuclear engineering. Diversity simply did not exist. The top dog at Halliburton 50 years ago, told me, “We need more nuclear engineers. It is just not simple.”
Now I read “There’s a Simple Answer to the AI Bias Conundrum: More Diversity.” Okay, easy to say. Why not try to deliver? Believe me if the US Navy, Halliburton, and a dumb pony like myself could not figure out how to equip a person with the skills and capabilities required to fool around with nuclear material, how will a group of technology wizards in Silicon Valley with oodles of cash just do what’s simple? The answer is, “It will take structural change, time, and an educational system that is similar to that which was provided a half century ago.”
The reality is that people without training, motivation, education, and incentives will not produce the content outputs at a scale needed to recalibrate those wondrous smart software knowledge spaces and probabilistic-centric smart software systems.
Here’s a passage from the write up which caught my attention:
Given the rapid race for profits and the tendrils of bias rooted in our digital libraries and lived experiences, it’s unlikely we’ll ever fully vanquish it from our AI innovation. But that can’t mean inaction or ignorance is acceptable. More diversity in STEM and more diversity of talent intimately involved in the AI process will undoubtedly mean more accurate, inclusive models — and that’s something we will all benefit from.
Okay, what’s the plan? Who is going to take the lead? What’s the timeline? Who will do the work to address the educational and psychological factors? Simple, right? Words, words, words.
Stephen E Arnold, July 25, 2024
Comments
-
- Subscribe to Beyond Search
Feature archive
News archive
-
