I made myself a promise.

I promised I would write everyday. At least 250 words on a topic related to Generative Artificial Intelligence, but not directly mentioned in any comment, post, article, newsletter, podcast, video or other content I had consumed the same day.

I'm considering tonight, 7 January 2024, the first "day", as I've had the flu every other day of this calendar year and am finally lucid enough to gather my thoughts and commit them to this medium.

Today's thought is about "nothingness". It is wise, and profound, and vast, and well-considered to reflect on the reality of the truism that something, i.e. all of us, and everything and everyone we know, came from "nothing", and are now, "something". And, when you are mindful, and have a moment, I truly encourage you to consider how truly wondrous that is.

But that is not the nothingness to which I refer. I cannot tell you, reader, how often I hear, read, watch (often in horror), observers, experts and everything in between act, critique, judge and invict (is that a word?), against Generative AI as if it is being used in a big vacuum of nothingness. Like we'll just apply GenAI in our schools, our churches, our companies, our governments and nothing, no one has ever come before, or will be needed hence. The worse offense is when people report the results of experiments, either their own personal experience (which is essentially an n=1 experiment) or a scaled experiment and then highlight all the bad things that GenAI doesn't do, shouldn't do, won't do or can't do today, but conveniently, in their Conclusions fail to mention that the verysame biases that they just, a moment ago, mentioned were an issue for GenAI are manifest, 10x more commonly in the humans in the same environment where GenAI will be a copilot. Or, that GenAI replaces a legacy computing system or software approach that was even worse and, while not perfect, that it is better and improving every day.

I'm not being pollyanna-ish, but there are multiple reasons scientific research in the rigorous technical disciplines has evolved to have robust designs such as comparator studies, placebo controls and randomizations. Part of those considerations are to minimize the biases OF THE EXPERIMENTERS, and part to ENSURE CREDIBILITY FOR THE IMPLEMENTERS so they know where and when to apply the findings.

Which brings me back to nothingness. Remember, we all came from nothing and are now something. When we seek to implement something new, anything, we must contextualize it; e.g. does GenAI have 3-27% hallucinations? Possibly, but how often does your colleague down the hall tell you with perfect accuracy the right answer to very technical questions in your domain of expertise? It's not just does something exist, it is does it exist in a context compared to other considerations of merit and what might that mean for my stakeholders, my environment, when all else is considered, e.g. cost, impact, engagement, etc.

Of course, this is just my personal opinion, not that of my employer; I could definitely be wrong, and I am always happy to hear thoughts and feedback.