As I looking at the feed of technology stories this week, two things jumped out at me.
First, Valentine's day happened to fall on the monthly release day for Microsoft updates and feature previews. If you work with Azure services, specifically in the data platform space, there are a lot of update announcements to work through from this past week.
Second, the constant feed of news about AI-driven chat platforms isn't going to abate any time soon.
Just over a week ago Microsoft announced the release of the new Bing with OpenAI technology in limited preview. Within 24 hours, Google announced their entry into the AI chat space with the Bard chatbot. Unfortunately for Google, Bard's announcement started with a factual error and the internet couldn't handle it. 😉
While the technology is fascinating and the signup numbers outpace nearly all previous "unicorn" projects, I have a confession to make.
I didn't try to use ChatGPT until earlier this week.
There are a few reasons why I waited so long, not the least of which was simply having the time to learn where to start. But if I'm being honest, it felt like a gimmick ready to waste more of my precious time. Within days of the public preview, Twitter, YouTube, and LinkedIn were (and still are) overrun by "10 Best Prompts for ChatGPT" threads and videos. I just didn't have the need or time to risk being sucked into a vortex.
As a database developer I could appreciate the samples of code writing and query generation. With the correct prompts it was fun to see ChatGPT provide digestible answers to technical design challenges. But in the end, it felt like everything I saw was another example of someone spending time trying to get ChatGPT to provide the answers that they could have coded in less time.
However, something finally pushed me over the edge on Monday.
My New Research Assistant
I was doing research for some blog posts about PostgreSQL patches and upcoming features. There is an endless stream of information for each topic that delves into the history, highlights, bugs, requests for improvement, and more. The PostgreSQL documentation alone is dense and time consuming to get through. I realized that ChatGPT might be able to help me summarize some of the questions I had.
I wanted to direct the service to a set of publicly available email threads and have it summarize the conversation. Unfortunately, that's not possible, yet.
Next, I tried to copy portions of the back and forth into ChatCPT but my text selection was always too long (with no feedback on what the maximum length was). Finally, I realized I had to take a different approach.
As I re-read the email thread conversation, I used ChatGPT to explain concepts that weren't clear to me. While I could have done the same in Google or another search engine, this did end up being a more natural experience. "What is the difference between X and Y?" "Is Y better in situations where Z happens?"
The answers helped me understand the concepts at the level I needed without clicking through multiple web pages or Stack Overflow threads.
Finally, I had found a use for ChatGPT - and I probably should have tried this sooner. 😊
Your Skills and Knowledge Are Still Valuable
While I came away with a sense intrigue, I also came away with a stronger sense of how essential it is to retain and encourage professionals that care about their craft. For the time-being, at least, our skills and knowledge are more valuable than ever... with or without ChatGPT.
Yes, ChatGPT might have the ability to quickly suggest the right SQL JOIN or which WINDOW function will produce the results you are expecting. But ChatGPT will never know your specific, private business data (and if it does, we have other problems to worry about).
You know your problem domain best. You know the distribution of your data best. You understand the nuances of how your data model works and the access patterns throughout each day, week, and season.
ChatGPT and all the AI assistant bots coming down the pike are just additional tools at your disposal. Every day we have a growing stream of new information coming at us and having a tool available that can quickly summarize and correlate that information will be particularly helpful.
Use these new tools wisely, not blindly.
What About You?
Have you found a particular use case for tools like AI chat that's benefited your day-to-day job responsibilities? How have tools like these helped you solve database or analysis problems?