I'm Sorry Dave, I Can't Let You Do That

TL;DR We’re still a very long way from Her or HAL9000…

There’s so much excitement and hype around the promise of AI. But how much reality is behind that promise today? While the shortcomings of current GenAI models are well documented, the hype cycle driving the AI frenzy seems to be strong enough (for now) to relegate even moderately contrarian views into the ‘you must be a Luddite’ bucket…

I’m not an expert in the field, which makes me a worthy litmus test(er) for applied AI. 

I found a fractional project that caught my eye, but the gatekeeper to the project wasn’t a team or a person – it was an online assessment tool that combined IQ test-like questions with AI chatbot interactions, both prompt engineering and conversational Q&A. Initially, the experience was fun - brought me back to college and the (good) stress of performing on the clock and under pressure. But…

As I progressed through the tool, the site froze & I was presented with a ‘white screen of death’ with an unending swirling circle. None of the typical browser tricks worked, and when I then switched to a different browser & logged back in, I got a prompt that my (incomplete) application had already been submitted. Wha?

Interestingly, and perhaps tellingly, there was no way to contact a human for support. Or even to submit a support ticket. Only an AI chatbot that was not particularly helpful. Or empathetic.

Who knows if I would have gotten the project anyway? The point is, I never got the chance to have the conversation because…BUGS! Everyone in tech knows there’s no such thing as bug-free software – whether you’re talking about code bugs or GenAI hallucinations. 

This experience won’t change my life, or even frankly, my interest & excitement for AI. But imagine if AI had been implemented this way in the cockpit of a 777. Or in a surgical robot in an operating room. Or in a weapons system on the battlefield. The outcome could be catastrophic.

The point here isn’t to throw cold water on AI but to provide an anecdote to remind us where we are.

To use a parenting analogy, AI is still a toddler that still requires us to care for and feed it, change its diapers, and teach it how to behave. You wouldn’t dream of putting your two year old behind the wheel of a car, so we should be mindful in the same way with our approach, adoption & deployment of AI at this point in the game.

PS - I’m not going to name the companies involved because my intent isn’t at all to disparage or criticize them. Also, I didn’t use AI to write this post…