We have an annual event at Computronix during our final all-company meeting of the year. It’s highly anticipated by every ‘troid (that’s short for Computroid, which is what we call ourselves) and easily the presentation with the most crowd interaction, laughter and comments. We all leave with a lot of questions, a list of things to research and best of all, a sense of community. What is this event called? ‘Jim’s Tech Rant’.
Jim den Otter, the CTO of our company, has been delivering these Tech Rants for the past 30 years. He describes how his Tech Rants have “morphed a bit over the years, it’s maybe less purely technical now, and more business and social impact related. But essentially, it remains a cross between rant, stand-up routine, industry gossip, and ‘what matters’ in tech. And it provides staff with the confidence that we’re always paying attention to what’s going on and what we think will matter. “
This year was no different, he spoke on the topic almost everyone is talking about…Artificial Intelligence. Namely, how AI will affect our daily lives, and how it will affect Computronix and the business community in a broader sense.
Because we all enjoy Jim’s Tech Rant so much, we thought it would be fun to let others in on what’s going on inside Jim’s head. We will be highlighting various aspects of his presentation in our next blog series. It’ll be interesting to look back years from now to see if he was on target and how much AI has changed. Only time will tell. Honestly, that is what is so fun about his Tech Rants, we get an inside look into the musings inside of Jim’s head and he doesn’t have to be right. He clarifies this point when he states that, “he gets to spout his own opinion about things because, after all, this is just a rant.”
“He began with an overview, stating that OpenAI developed a natural language processing model called GPT-3 (short for Generative Pre-trained Transformer 3). This model is designed to respond to natural language questions by applying its knowledge and providing relevant information.”
“However, it has its limitations. Since it was only fed information up until 2021, it doesn’t possess any knowledge beyond that year. Consequently, it is susceptible to confusion. You can feed it false information, and it will process it like any other.”
Next, Jim explains how he asked it, “from Microsoft’s Northwind database, create a chart that shows revenue by quarter for the last year. And it responded with, ‘I’m a text-based AI, I can’t give you a chart, but I can give you a query that can give you the data to create the chart.’ And it did! It spits out a perfectly acceptable SQL query against the database that would, in fact, give me totals of revenue by region for the last year.”
Eventually, and with a few more instructions, asking it for some Excel VBA scripts and ASCI II art, there was a chart in Excel, right after saying it couldn’t produce a chart in Excel. He continues with a process for creating reports and how we will utilize this in the very near future.
Then comes the transition with a good question, “What are some of the problems with Chat GPT?“. And since this blog post is starting to get a bit lengthy already, we will have to pick up right there in part two, discovering what Jim sees as some of the issues with AI.