Wednesday, April 3, 2024

Tricks, Tools, and Transparency

When it comes to using AI, honesty is the best policy

By André van Heerden, Communications Director

No one wants to be tricked, and no one wants others to have an unfair advantage. Some of the growing concern regarding AI technology comes from these feelings.

Everyone wants a level playing field, but at what point are we measuring tools versus measuring human performance? Is there a way to determine which uses of AI are acceptable and which are not?

I remember back in the early 1980s when professional tennis players began to use racquets made of other materials besides wood. In Major League Baseball, only wooden bats are allowed, arguably for traditional and performance reasons. Why would tennis be different?

My father argued that the new racquets gave players too much power, and weaker players with a big serve could now beat better players. But over time, since every player could choose what racquet they wanted, everyone had the same advantages.

Within auto racing, there are a myriad of very detailed restrictions on what cars can or cannot use. And these rules differ for F1, NASCAR, stock cars, and many more.

Electric vehicles are making these debates even more complicated. Many would argue that auto racing has become more about the technology than about the actual drivers.

Is that what race car fans want? Is it about the driver or the machine or the technical crews? Or is it about all of that within closely monitored rules?

It looks like new legislation in Ontario will require job postings to disclose if workers’ job applications will be reviewed using AI technology. Why would this matter? Clearly, AI programs can be told to look for certain key words and credentials within someone’s resume.

But is that a fair assessment of the candidate’s application? And what if someone used AI to create their resume? Would someone need to disclose that as well?

You could have one AI system evaluating the work produced by another AI system. Would it be better to submit a different resume if you knew it was being reviewed by a person rather than AI? And how impartial or fair would an AI system be?

Some might assume that a computer system would be more fair in assessing applications than a person, but those computer systems are still designed and created by people and therefore can carry significant biases. The recent news coverage regarding Google’s Gemini chatbot certainly reveals that AI systems can be just as prejudiced as people.

I’ve spoken with a number of high school teachers and postsecondary professors, and many believe that any true test of a student’s academic abilities will have to be done with only a pen and paper and the absence of all technology. But a student may question if the test is relevant to preparing them for the future.

Is it better that a student know how to spell or how to properly use AI? This guide of how to ethically use AI for University students is informative and helpful.

Disclosure of all sources seems to be one possible solution. If someone used AI to design a website, they should credit it. If someone used AI to generate a business plan, they should disclose it.

But how much would we need to credit? Is getting directions to a workplace use of AI? In the tennis analogy, would players need to reference what high-performance shoes they’re wearing or how light or breathable their playing clothes are?

When it comes to AI, honesty is the best policy for at least three reasons:

By being open about how much AI was used, it gives the audience a better understanding of the author’s work.
Disclosure of what AI tools were used could help others to learn how to use those same tools.
If certain tools perform better than others, that should become apparent.

For the record, my tennis racket is made of graphite and the preceding blog was written using spell check and Google search.

Tricks, Tools, and Transparency

You might be interested in