The cool kids like talking ChatGPT - AI's new sWeET dream?

Every second post in my feed is about ChatGPT - how it's taking over developer jobs and will replace all things human. Well... not really! There are tons of useful applications of ChatGPT especially in the service industry - first-line of support, connecting people to the right experts, task automation, etc.

ChatGPT was definitely not built over night, and it's not a complete system (AI is never complete). It takes many hard computing and natural language problems over the years and condenses it into one interface. Solutions to natural language generation, text summarisation, question & answering are some that feed a system like ChatGPT, and of course the more knowledge it is exposed to, the more the people use it, the better it gets - just like the human brain!

I haven't really followed the recent developments in this area of AI, but just the buzz wanted me to give it a try - and the first question I had like any business would in making that investment of time/effort/money - what's in it for me? Would this help me in my day-to-day work - ROI being saves me time to invest in solving more critical problems that my customers face. Can it do something that I can't - ROI being better experience for my customers.

So I decided to put this to test. I signed up on ChatGPT (pretty slick interface) and keyed in a technical question that a prospect had regarding a specific feature in the product and how it could work.

First look, pretty impressive I must say - even I (someone whose job is to work with the product day-in day-out) was convinced by the response. Now, if I were a prospect I would take it that this either works out-of-the-box or will work with some minimal configurations. Reality check, solutions engineering is not about providing a one-size-fits-all solution and often prospects/customers need to weigh in on the pros/cons amongst multiple options and decide which works best for them.

Diving deeper on this response, and several discussions with a lot of superstars at Confluent (more experienced SEs, product managers, ...) together with the prospect we decided white-boarding the various combinations and do some testing (which means actual hands-on configuration and wiring things together) of the solutions that came up from there. (Psst... I told you ChatGPT can't replace what you do... at least not yet)!

So in summary:

• ChatGPT does not replace work done by developers/engineers. It will evolve to removing repetitive/redundant conversations and make things more factual over time. ROI - saves time/effort in turn $$$ for the business.
• Processes and workflows will prove to be more critical, and tools like ChatGPT will assist in progressing some of these tasks faster - ROI being increase in productivity of the knowledge worker.
• There are a lot of good technologies out there that reduce the time/effort one puts in into their day-to-day tasks. Let's appreciate the ones that dare to go open-source! There is economics on the other hand that decides if people want to save time, money or effort, and put it to better use and if so, what is that better use.
• As Noam Chomsky rightly puts it - the grave side of such technology could be that it takes away the joy in learning. Did I enjoy the process of discovering and trying out all the above solutions before accepting/rejecting the response from ChatGPT - absolutely! ROI - I learnt a bunch of things and had some good conversations with very knowledgeable folks.