Readers who have somehow escaped learning about ChatGPT and other AI-based chatbots should read a new book by my longtime freelancing colleague Pam Baker, “ChatGPT for Dummies.”

Baker has been covering various AI-related topics for nearly a decade, including writing several books on machine learning and data science. She has a long business-to-business tech resume, having written for dozens of publications over many decades, along with being an instructor for LinkedIn Learning.

Her latest book is full of useful information for both neophytes and experienced pros alike, along with Baker’s insights into the coming future of our robotic overlords. (Only somewhat kidding here.)

Given the latest news that ChatGPT for Enterprise is now out, this book comes as a propitious time. “Enterprises are going to struggle with the spread of AI and try to control it,” she told me in a recent phone interview.

About those future overlords: As she says in her book, “The threat to your job isn’t ChatGPT but the people using ChatGPT and other AI tools.” That is true. She recommends that information technology workers should “start thinking of and creating new ways to use ChatGPT to your advantage. Develop prompt engineering skills, pay attention to how it is shaping and redefining tasks, actions, jobs and industries over time.”

Germane to those prompt construction skills, she provides lots of tips and links to other resources on how to improve your prompt prowess, including thinking like a storyteller, not an inquisitor, and focusing on prompt chains, or using multiple prompts to focus your query. With all the stories on chatbots, this advice seems unique and well-thought-out.  She told me that “journalists know how to phrase a prompt, and the better the prompt, with sufficient context and background, the better the response from the chatbot.”

Chatbot trends

She identified several trends in her book. First, whether chatbots will replace general search engines isn’t likely, but she says that there will be a slow death spiral of keyword jiggering and search engine optimization professionals. In our call, she said that “professional researchers are going to continue to rely on search engines, because of chatbot disinformation and other problems.”

But she said the general public is going to find AI models “much easier and quicker and more satisfactory because they are more conversational. SEO is aimed at getting eyeballs and moving products. If and when the consumers leave the search engines, SEO won’t work very well. Until the AI models get more trustworthy, few people will use them.” In her book, she mentions that “ChatGPT is a convenience store offering a one-stop shopping experience for the internet. It is a pattern detection software mirror.”

Next, there is a difference between an app that uses AI features and an AI program that can incorporate apps and add features programmatically, such as through the use of plug-ins. “This is like the difference between a first grader and a junior developer writing some code,” she told me. The former is just a lot less powerful than the latter, and she expects to see massive innovations on both fronts.

One chapter in her book is devoted to this topic, with plenty of examples for readers to experiment with, such as the Ghostwriter add-on to Microsoft Office. And she mentions other companies that have incorporated the tools into their products, such as Snap Inc. and ChatPDF.com, which extracts content from PDFs, and AutoGPT, designed for internet searching for products, which could be used as a personal shopper.

All of these innovations mean that chatbots will be tremendously disruptive for IT managers. In her book, she mentions quite a few of them, including how knowledge work will be augmented, such as freeing data analysts to look deeper into their data. Marketers will produce more content. There will also be more superapps, or those that automatically select their own plug-ins to augment their performance.

Finally, there is the matter of the actual intelligence, perceived or real, of the chatbot tools. She discusses the various metrics that are used, showing that depending on which one is chosen can mean the difference in how a chatbot is perceived. “The simpler things tend to trip up an AI, such as failing sixth-grade math but getting more complex questions right,” she writes. She doesn’t mention this example, but the researchers who gave ChatGPT the Wharton MBA exam earlier this year is a good case in point (the AI was graded a B).

My biggest beef with her book is the poor quality of the screen captures and other illustrations that make them very hard to read. Baker realizes that keeping her book current in such a quickly moving field can be a challenge, and she has signed up with LinkedIn Learning to teach several AI-related virtual courses that will be available this fall and will be a mechanism to keep her content fresh and up to date.

The first one will be devoted to using ChatGPT for help desk applications. As Lori MacVittie, F5’s distinguished engineer says, “AI is game changer, but that game is chess. It’s about strategic use and making sure you have a scalable and secure foundation on which to build future capabilities.”

Image: Adobe Stock; photo: Pam Baker