I think I may be getting way ahead of myself with all this AI (Artificial Intelligence) stuff. Let’s take a step out of the rabbit hole for a few minutes.

When was the last time I ever wrote about climbing out of the infamous rabbit hole?

Let’s look at a large-scale attempted use of AI with a focus on automation. There was a study of over 2,000 mystery medical cases. The research looked at all files from 3 different perspectives. First, doctors diagnosed on their own, then 5 of the best and latest AI models were used, and lastly, the doctors gave their final diagnosis after reviewing the AI suggestions.

The results: The doctors’ reviews of the AI suggestions before giving their final diagnosis were right about 85% of the time, more often than AI alone, which was about 17%. 

This is a research project exploring automation vs. collaboration.

The article I read was extensive, with many examples exploring this binary choice.

But sitting in the sunlight beside my rabbit hole, I see other choices, rather than just 2 as suitable goals, at least for the next decade or so.

Instead of aiming to make humans into bystanders to AI reading x-rays, for example, we could move to a collaboration style with medical personnel.

Let’s move to a realm I am more familiar with, driving my vehicle, which I park near my home every day.

There are enough examples of 100% self-driving cars. On highways, with gentle curves and all necessary white lines clearly visible, in other words, ideal conditions, I could cautiously support 100% self-driving cars. But only partially.

Here comes the what abouts.

What about a long stretch of newly paved asphalt before the lines have been painted?

What about a detour into a formerly head-on driving situation as my 2 lanes take a detour around a bridge repair?

What about safety features in newer cars to give a warning about lane drifting, for example? Keep this AI or stop it now?

I won’t go on catastrophically anymore.

Self-driving currently works as long as humans can take over quickly with situational awareness. I’m not confident I could do that if I had to be awakened from a deep sleep.

So, designing AI to collaborate with humans to different levels depending on the seriousness of the moment would be a worthwhile goal.

I still take the collaborative approach with my Blogs. I use a spell checker and a grammar checker. But I always proofread the suggestions. I never click on ‘accept all suggested changes.’ I collaborate with the AI software even in this low-risk consequence situation. I’m a better writer for the help I get, and I’m confident my readers are unknowingly appreciative of the care I take.

I now have the language with the help of the title of this Blog, to be supportive of AI but not blindly so.

Please give this a bit of a think. Where have you found AI helpful in your life, and where do you think we’ve trusted it too much?

I am curious about your thoughts. Please comment below or send me an email with your bit of a think.

My thanks to The Atlantic magazine, which inspired the topic.

Photo by Yivan on Unsplash

If you enjoyed The Blog, please share it with others. Thanks.

And my thanks to St. Albert’s 50+ Activity Centre for making this Blog possible.

Volunteer Blogger

glenn.walmsley@icloud.com