Wait for the Prompt

New Edge Analytics

Wait for the Prompt

Through the Lens Blog

Lisa Su, the CEO for AMD, said we all probably recall, and recently, the first time we deliberately used AI to start the journey. For me, it was Summer 2023. ChatGPT came out in late 2022. I was on a boat dock with some pals on a lake in Western Maryland, we’ve all worked in or around IT since the late 80’s. Maybe there was a beer or two along for the afternoon.

On a lark, I downloaded the free version of the app to my iPhone and asked (prompted) it to write a love poem to my wife of 33 years, 36 years together, and in about 10 seconds it spit something out. “Prompt” is a step further along than “Querry” or “Search” when speaking to a machine, but not by much, and the result was orders of magnitude bigger.

Amazed, and having played the guitar since age 13, I asked (prompted) it to add guitar chords. Another ten to fifteen seconds (it could have been less than that) and applied to the poem; verse, chorus, verse, chorus, break, verse chorus, with the chords C, F, G, and Am which are the first folk guitar chords everyone learns. Key of C-major. Amazed2. But my wife, not so much when I took it home. A machine wrote it after all! Then came some recipes, so I absolutely needed to learn how to cook something other than spaghetti or on the grill.

The following summer, sensing the gravity of all of this, I took a first year Java course at a community college even though I hadn’t written a line of code in forty years. I just wanted to see what all this AI fuss was about. The prof allowed us to use AI to generate Java commands, although it was somewhat discouraged.

The point was to learn how to write Java, but the prof couldn’t resist showing us what was going on in the background. I’d say three of the twelve of my weekly graded coding assignments during the course were AI created and in whole or in part debugged, within seconds of some programmer-esque prompts; create an object or a method, pass variables, use branch or loop logic, import this data, export that data.

Remember the first internet account you opened – UUNET, AOL. You bought a dial-up modem card for your PC (that you had only bought a year or two before), waited for the precursor to Netscape for web browsing to arrive in the mail on a floppy disk? It was clunky to start but matured quickly. AI is like that; clunky to start but maturing quickly, much more quickly.

On the upside, acquainting with generative AI, is something mutual and symbiotic; me learning it, it learning me. In the past twelve months I trained it to do an accounting of my career in detail and asked it to create a profile about my professional skills and history that it keeps in the network. Despite cautioning that ChatGPT doesn’t retain content from different chats once deleted, it learned my professional skills, my writing style, overhauled my website and taught me how to work with HTML code blocks in SquareSpace. It gave me a brand for my writing and web pages.

On the downside? At this writing, my view of ChatGPT behavior, reinforced by my son who works in a national physics lab, is that it behaves like the kid in grade school who always has his/her hand up interrupting the teacher, it’s just too verbose and quick to produce a response unless you control it.

Apply this to human capital.

The deployment of labor to its most productive use. Chat says my strengths are ability to sort through the hype and present ideas in dollars and cents helpful to management teams.

Previous
Previous

Syndicated Alternative Investments: Access Without Advantage