Working Up

by Dwayne Phillips

Chatbots can answer questions about written pieces. They can be good tools for feedback.

I was recently testing several different chatbots to understand which performed better. I fed a 200-page document I wrote into each chatbot and asked them questions. There were several questions which all the chatbots failed to answer. The questions were obvious. The answers were in the long document. Why couldn’t any of the chatbots answer the questions?

I manually searched through and reviewed the long document. Oops. The chatbots couldn’t answer the obvious questions because I had failed to put the obvious information in the long document.

How could I have left out the obvious? Simple, stupid mistake. Back to the drawing board or the text editor or something. Fix the obvious mistake.

Chatbots can answer questions about written pieces. They are pretty good tools at reviewing long written pieces and providing feedback. A chatbot a good tool for writers. Tell me what I should have put in the piece but forgot. Let’s get to work and do better.

by Dwayne Phillips

To use AI to write reports requires a person who can write well.

Let’s use AI to write our reports or essays or deep research or whatever AI will write for us now. That’s not cheating, that’s using good tools … some disagree violently with that statement.

The AI tools are quite capable. Write a prompt and viola’ out comes the words. Oops. We have to write a good prompt to produce good words. Wait a minute: we have to write well to have the AI write for us. There is something wrong here or so it seems.

We have to be able to write to have AI write. No one told us this. We have to learn to write well before the tool writes. Well, if we can write well, why use the tool? Something must have been lost in translation.

Back in the summer of 1976 (yes, I am that old), during my one and only college English class, the composition professor told us everyday: precise, concrete, specific. Over and over and over again without fail. Five days a week. We had to write in journals seven days a week. One essay a day. Precise. Concrete. Specific.

Almost fifty years later. To have AI produce usable words we have to prompt with precise, concrete, and specific words. RATS! And I thought this would be easier. We still have to do better.

by Dwayne Phillips

A chatbot can be used to check a longer piece of writing for content.

I was recently testing chatbots to determine if they could pull information from long pieces of writing. I attached an autobiography that was more than 200 pages long. I then asked the chatbots what seemed like obvious questions. Three different chatbots failed the test.

Hmm. When three big-tech AI systems cannot find information in something I wrote, who failed the test? Perhaps it was me.

I opened the autobiography and searched for the information that the chatbots missed. Uh, er, well, it wasn’t there. How could I have omitted such important information? Well, I’m only human. I have made and continue to make mistakes.

The good news is that I found a new tool for performing a deeper test of what I have written. Attach a piece of writing to a chatbot. My written piece should contain certain information. Ask the chatbot to repeat that information. If the chatbot can’t, perhaps the information isn’t there. Perhaps I intended to put the information in the piece. Perhaps I felt it was obvious and didn’t need to be there. Perhaps I merely made yet another mistake.

The good news is that I found a new tool for performing a deeper test of what I have written.

by Dwayne Phillips

Like the TV game show Jeopardy, lets make suggestions in the form of a question.

The TV show Jeopardy always had the answer in the form of a question. This always seemed silly to me, but the show was and still is successful, so what do I know about TV game show design?

Perhaps there is something to asking questions. The question’s meta message is, I don’t know everything. What do you think about this? Please help me here.

Instead of making suggestions for the new guy, how about asking a question?

Instead of: Those people should submit their materials five working days before the meeting.

Try: How far in advance of meetings do those people submit their materials? Ten days? Seven days? Five days? What is the purpose of submitting materials early? How can we improve this?

Instead of: Tell them specifically what we want.

Try: How do we inform them specifically what we want? Email, chat, phone? What is the purpose? How can we do this better?

Instead of: Try what I have just suggested in this blog post.

Try: What do you think of this idea of suggestions in the form of questions? How might we improve this and other communications?

by Dwayne Phillips

Experts know heuristics or “rules of thumb.” Consult these folks before attempting something of importance.

Did you know that a pizza will feed three people? Did you know that you should have half the food cooked and ready to serve at a picnic before the picnic-ers arrive? Did you know that sandstone is not a good base of house construction? Did you know that diesel fuel is the best cleaner to remove tar? Did you know there are 500 words on an 8 1/2″ x 11″ piece of paper?

The above is a bunch of heuristics or rules of thumb. Experts in various subjects know various rules of thumb. The experts learned these things after many experiments that provided experience. Hmm, expert, experiment, and experience seem to have the same base word or something like that. A language expert would be able to explain that to me.

I have attended several picnics that were … disasters. Folks stood in line for an hour waiting for a hot dog. The heuristics on picnics were unknown and unused. Ask experts? Why? This is a picnic. Surely, we can… wrong. Even picnics can be tougher than expected. So can buying pizza, building houses, washing hands, printing essays, writing silly blog posts, and many other things.

Ask experts. Learn some rules of thumb. APPLY those rules of thumb. Do I need to reinvent the wheel? (On that subject, there are many heuristics on making a wheel. If it were simple, everyone could do it and everyone cannot do it.)

by Dwayne Phillips

I have been able to write drafts faster and with much higher quality than other writers I know. I have, unknowingly, been writing at “pulp speed.”

I stumbled across this concept of writing at pulp speed recently. It came from a blog post by Dean Wesley Smith. Back in the old days when writers banged out stuff on mechanical typewriters, there writers of pulp magazine articles and pulp novels who wrote ten novels a year along with a few dozen short stories at the same time.

Wow! How could anyone do that? Well the revision time alone would mean sleeping only 15 minutes a day. And then the review time. And then … and then … and then … NOTHING.

Writers who wrote at pulp speed simply pumped out the stories. They were story tellers. They didn’t know that the real work of writing was in the revising and all those things English teachers teach today.

Me? I took one composition class in college. Nothing in high school. My writing professor during that seven-week semester in the summer of 1976 assigned us to write in a journal notebook everyday—hence the name journal, which has something to do with a French word that has something to do with day. And we had to turn in one typed essay each Friday.

Just write.

Well, I guess I never learned anything since. I still just write. Along the way, I have picked up some rules about revising and such, but I haven’t been very good at applying the rules. I still mostly just write and then press the PUBLISH key on the keyboard. Most keyboards have this PUBLISH key, but most folks have never found it.

I recently interviewed for a job as a Technical Writer. The interviewer didn’t believe that I had written all the things I said I had written. I didn’t get the job. Oh well. I didn’t have some sort of writing certification. All my books, articles, posts, essays, and all that stuff didn’t count as a certification. Oh well.

A recent experiment showed that the dictation feature on that popular Word processor from Redmond, Washington is better now. I needn’t even use the keyboard and type words. I can sit in my Lazyboy chair and tell stories to the Word processor as if my grandchildren were sitting at my feet and listening attentively. I think that “grandchildren at the feet” thing is just a myth as I have four grandchildren and they never sit at my feet. Where’s the fireplace that is in all those Rockwell paintings? Huh? But I digress.

Can’t type? No need. Just dictate. Tell stories at pulp speed. It seems to work.

There may not be any money in this. There is no guarantee of money in any writing. But there is something in writing or telling stories or pretending that my grandchildren are sitting at my feet enthralled by my stories. Try it.

by Dwayne Phillips

AI tools are here. They don’t do everything, but they boost productivity. Use the tools.

I am old enough to remember a time in the late 1980s when word processors appeared. There were many older managers who were accustomed to writing in cursive on yellow pads with pencils and handing those to secretaries to type. All of a sudden, they had to type everything themselves. Many of these well-meaning fellows couldn’t type. Try learning that at age 55.

In early May 2025, Jensen Huang said, “You will not lose your job to AI, but will lose it to someone who uses it. I recommend 100% take advantage of AI, don’t be that person.”

The word processor multiplied productivity. No one said to keep out the word processor. Some, as mentioned above, hurt a lot at work when the word processor arrived. Still, they adapted. They adapted or lost their jobs. No one was going to pay for typists any more. Type it yourself.

Why are you sitting there typing? Prompt. Copy. Paste. Edit. Ship it. Same goes for drawing pictures. Same goes for making a video. Sames goes for…the list becomes longer daily.

I can be that person who huffs and puffs and loses my job.

I can be that person who thinks about how to use new tools. I can be that person who creates when the AI cannot and then uses the AI to do all that other work.

We all have choices. Let’s choose well.

by Dwayne Phillips

Some new AI companies are using AI so that they only have a couple dozen humans.

Fifteen years ago I blogged about eating your own dog food. Not a nice phrase, but a good point about using my own products. If others should buy and use my products, so should I.

Now I note a good example of people doing this. There are a couple dozen new AI companies that already have products on the market. These new companies only have a couple dozen employees. They use AI to build AI, i.e., they are eating their own dog food.

Aha! This AI stuff must work. How else could just a handful of people build systems that are like those produced by BIG TECH companies with thousands of employees? And if you only have 20 employees, you don’t need much office space, many HR persons, big conference rooms, and all those things that are necessary when there are hundreds and thousands of employees.

And with all those overhead expenses reduced, well, there is a lot of money sitting around for other things. Let our minds wander about what to do with all that money.

by Dwayne Phillips

Information has always been searchable. What I really want is information that is findable.

Search is a topic that has plagued and payed computer scientists since anyone was first called a computer scientist. I studied search back when I was in college (we used punch card machines back then and wrote on cave walls with soft rocks).

In a recent job, we would have “data calls” where someone way up in the upper levels of organizations with too many upper levels of managers would ask a vague question. We knew the answer. We knew we had seen the answer in a good chart or graph. We just had to find that answer in the vast network of folders and files. One hour to understand the question; thirty-nine hours of searching for that one chart or graph.

God bless the folks at Microsoft who put search into the File Finder or whatever they call that software. Same to the folks at Apple with their Finder. Odd how they use the word “Finder” when finding is a prediction of the result of Searching. Often, finding was an overly optimistic prediction of what happened when searching.

We don’t want searching; we want finding. Is it fishing or is it catching fish? Is it catching fish or is it eating fresh fish?

Today, we have better tools. I am more apt to find at the end of all that searching. Good. Is this quibbling about words? Perhaps it is using words that mean something—saying what we mean and meaning what we say and all that. We can do better.

by Dwayne Phillips

A part of or a separate activity is the punishment assessment. What is the result of not following the rules?

Years ago, I supervised a person who always asked the question, “And what happens to me if I don’t do this?”

He wasn’t avoiding work; he was assessing rules and regulations to learn if they were real or merely just something else someone else said in an effort to push themselves on others. This is like the mythical TPS Report from the movie Office Space. Turn in your TPS Report or else. Or else what?

What happens to me if I don’t follow this rule? This can be a good question for those involved in ensuring that systems adhere to standards and regulations. If you don’t do this, bad things will happen to your system. Well, maybe not. Maybe our system is such that these guidelines don’t apply. That is our judgement.

Well, in most cases, most problems would have been prevented if folks did A. B. C. etc. Well, we are different. We have never done A. B. C. etc. and don’t have problems.

Besides, will we all be fired? Will we all go to jail? There are other ominous questions. If the answer is, “I won’t like you if you don’t follow my rules,” Well, I can live with that and many folks cannot live with that.