This past April, I was invited to speak at several conferences on the use of generative artificial intelligence, which I will call AI in this article. While the topic is outside of my usual leadership swim lane, I feel qualified to speak on it because I am an engineer and we are using it in the company.
I love new technology, especially when I have a reason to use it. Having a problem to solve with a new software solution is a great way to learn how to use the new tool, and how to push it hard its limits.
I remember the day way back many decades when the IT group rolled a cart into my office with my new desktop computer and followed it up an hour or so later with boxes of software. It was like being in a candy store. So many boxes. So many choices of capability. And no parent standing over you saying, “leave some for someone else.”
One of the boxes I selected was Lotus 123. (Yes, that does date me. For those of you who haven’t heard of Lotus, it is the precursor to Excel). I had been working on a “time to go from off to functional” challenge for the submarine combat system I was a systems engineer for. We had recently undergone a major hardware re-architecture, and the lead software designer’s plan for getting power on test status information and then loading software serially was not going to make our requirement for a fast load to an operational capability. We had engaged in several circular arguments, and this box with the software to easily show how long a serial design would take versus a parallel design was just what I needed.
The next day at the trouble report meeting, I slapped the glassine slide printed with my spreadsheet results on the overhead projector (yes, it was back in that era) and within about a minute the lead engineer mandated a parallel load. Case closed.
The right tool makes the job easier. In carpentry as well as engineering.
So, is AI the right tool for these times?
For people who have not been paying attention, AI has been embedded within our life framework for quite a while now. The original algorithms were developed almost three decades ago. We just didn’t have the memory, connectivity and compute power to develop them at that point. However, as technology has grown stronger, so have the AI capabilities.
Our Siri and Alexa assistants we have seen grow from amusements to a tool for gathering information needed when we couldn’t do it ourselves. I routinely ask Siri for directions, to set timers, to define something, and to settle a debate with information it finds.
And now we have large language models that make it even easier to communicate with the underlaying AI engines to do amazing things.
But it’s good to remember that these are emerging technologies.
Emerging. Learning. Essentially, the world is being used right now as a beta test site for these AI tools.
AI tools are built into our browsers. Think Google Gemini and Microsoft’s CoPilot. There are freeware versions of others, like ChatGPT. And many more, some with specialized functionality to be able to do tasks like mine large data.
These tools are not going away. In fact, they are going to grow in complexity and functionality over time. We are being encouraged to develop a relationship with our AI bots and one of the jobs I’ve seen listed as a ‘new tech job’ is AI personality developer. As someone who still misses her DOS command line, I’m not sure how that is going to change my life, but when it shows up, I’ll jump in and make a new AI friend. Or group of friends.
So, what do you need to watch out for when using AI?
First, remember that we are in generative AI’s toddler years. Toddlers get things wrong. So, if you have asked your generative AI tool for information, it’s a good idea to use a second and maybe even a third tool to answer the same prompt and verify the answer. Even better is to ask for the sources of the information to be cited so you can go directly to the source and see if what has been shared with you is accurate, exists, or if it has been taken out of context.
I read in today’s New York Times an article from Curtis Sittenfeld1 about his challenge to write a 1000-word beach story and compare it to what ChatGPT could write. The author’s work is known to have been used to train the AI engines, so ChatGPT was directed to write in the style of Curtis Sittenfeld. Both Curtis and ChatGPT used the same prompt concepts: beach read, 1000 words, lust, regret, kissing, middle age and flip flops.
The result: Curtis wrote a cute story that I enjoyed reading. The ChatGPT story was boring. I didn’t make it past the first few paragraphs. It just wasn’t engaging.
It was a toddler’s writing.
So, Curtis’ job as a writer is safe. For now. That toddler will learn over time. And students who use generative AI to write stories and papers and the like will pay the price because generative AI is just not that good at writing. Yet.
I tried to use ChatGPT a year ago to write a poem for a friend’s 70th birthday. I attributed the poor results to my lack of experience as a ‘prompt engineer’. Prompt engineering is the name for crafting requests to generative AI to get what you want from them. It takes a while to develop this skill, and every AI tool is a little different in what they respond to. Oh, and it pays to give them feedback, like “This did not include her name,” or “I asked for a four-line rhyming poem, and you gave me a six-line poem that does not rhyme well.” While the poems were not impressive, I did print out the prompt log and results, which was a very engaging read for the birthday girl.
Over time these tools will improve, and we will grow our skills to engage with them more efficiently and effectively.
Already I use AI for ideas, for video creation, for web searching. It has improved my access and cut down the time it takes me to find specific content and to create animations. I’m looking forward to what’s next and hoping I’ll stay in the field long enough to develop a relationship, maybe even a friendship, with my AI co-worker. And maybe when I am no longer working, and maybe not so functional anymore, my generative AI friend will take the time to remind me about all the good I did in the world while I was an engaged and active member on the bleeding edge of technology.
1Sittenfeld, C. (2024, August 20). An Experiment in Lust, Regret and Kissing. The New York Times. Retrieved from: https://www.nytimes.com/2024/08/20/opinion/beach-read-ai.html?campaign_id=2&emc=edit_th_20240821&instance_id=132238&nl=today%27s-headlines®i_id=48099420&segment_id=175686&te=1&user_id=2c68d81822512eeb33c3d2f792d61315
