Keeping clients' computers safe and profitable for over 30 years
Home Forms About Current Newsletter subscribe
Search All Articles
Browse by Category
Artificial intelligence, Chatbots, AI Writing, AI Search
What's it mean to you?
Preview:
Is AI (artificial intelligence) the next electricity or the end of the world? How can you experiment with it? What can it do for you now? What are its benefits and problems? All this is in one short article.
Is AI (artificial intelligence) the next electricity or the end of the world? How can you experiment with it? What can it do for you now? What are its benefits and problems? All this is in one short article.
Last month I wrote an article on AI art and described how I use it. For example, I did the picture above for this article with Playgroundai. This month I'll delve into AI writing and search.
A Couple Odd Samples
I normally use ChatBots as search enhancements, and will discuss that. But hopefully, to catch your attention early, I'll try and make the bots be creative and give you examples of my interactions with Youchat and ChatGPT. I'll follow it with more useful examples. But these examples give you some idea of where this AI stuff is headed.
Me:Tell me a very short story with a moral about a hummingbird on trial for eating a worm when starving. His judge is a cat.
Youchat wrote:
Once upon a time, there was a tiny hummingbird who was very hungry. He spotted a delicious-looking worm and, without thinking, ate it. Unfortunately, the worm belonged to a cat, who was not pleased. The cat decided to put the hummingbird on trial for his crime.
At the trial, the hummingbird pleaded that he had been very hungry and had not meant any harm, but the cat was still determined to find him guilty. However, just as the cat was about to hand down a guilty verdict, an old owl spoke up.
"I believe this hummingbird has learned his lesson," the owl said. "He was very hungry and did something wrong, but he is sorry and has realized his mistake. I believe that this is punishment enough."
The cat was impressed by the owl's words and decided to be merciful. He declared the hummingbird innocent and let him go, with a warning not to do something like that again.
The moral of the story is that we should all be forgiving, even when we have been wronged. We should be willing to accept contrition and give a second chance to those who have made a mistake.
ChatGPT Wrote:
Once upon a time, there was a hummingbird named Hummy who was on trial for eating a worm. The judge was a cat named Judge Whiskers. Hummy was starving and had been struggling to find food for days. He had stumbled upon the worm and, in his desperation, ate it.
The cat, who had never gone hungry a day in his life, was harsh in his judgment of Hummy. But, just as he was about to pronounce sentence, an ant who had witnessed the event spoke up in Hummy's defense. The ant explained that the hummingbird had been in dire straits and that the worm was his only means of survival.
Judge Whiskers, realizing the gravity of the situation, declared Hummy not guilty and even offered to share his own food with the little bird.
Moral: Necessity knows no law. And empathy and understanding can change our perspective.
After that answer, I followed up with ChatGPT and asked: but isn't it against the laws of nature for hummingbirds to eat worms? They're supposed to suck flowers.
It answered:
Yes, it is not typical for hummingbirds to eat worms as their primary diet consists of nectar from flowers and small insects. However, in certain circumstances, such as when food is scarce, animals may have to adapt and make do with what is available to them in order to survive. In this case, the moral of the story is about understanding and empathy towards those who are struggling and to not be too quick to judge, as we might not know the full circumstances of their actions.
I ask: Tell me a joke about a spider and an octopus who walk into a bar.
ChatGPT Wrote:
Why did the spider and the octopus walk into a bar? To see their friend, the bartender, who said, "What is this, some kind of joke?"
Youchat wrote:
A spider and an octopus walk into a bar and the bartender asks, "What can I get you guys?" The spider replies, "A beer for me and a glass of water for my friend here." The bartender looks confused and says, "But why does the octopus need a glass of water?" The spider replies, "Because he's drinking it through a straw!"
I said, That wasn't funny. Try again
It wasn't any better.
So I asked:
Are you likely to develop a better sense of humor soon?
to which it replied:
I'm constantly learning and improving, so I'm likely to become better at generating humorous responses with time. Right now, I'm able to generate text based on the input I receive and may be able to provide more lighthearted and humorous responses in the future.
I didn't know that free chatbots could perform like this. However, that's not an important usage yet, just a catchy one.
Overview of Usage Areas
Here are some areas AI writing is performing amazingly.
- Search: Most of my searches are questions like what does this error message mean? Or who is someone? Or who said something? Or what is something? For these types of searches, I don't use a normal search. I use either the Chat in You.com or ChatGPT. Both are in beta but both work better for me than a normal search when they are working. Both can be full or not working sometimes. Their knowledge base is also smaller than a general search. This is the most useful function I've found for AI writing.
- More in depth writing. These chatbots will produce students' book reports and essays. The New York City Department of Education has blocked OpenAI’s ChatGPT service on its networks, fearing that students will cheat on assignments.
- They can produce ideas and outlines for articles. You can ask it for an outline about a topic, or an apology letter, or wedding vows.
- They can produce short program code segments to solve particular programming problems. Apparently, they can also write malware for those miscreants incapable of doing it themselves!
- It can summarize complicated text clearly.
- They are being used as AI friends and counselors. I've never tried these, but given my assessment of counseling, psychology and psychiatry, it's probably a step up.
- As a help for people with poor English skills, so they write like an educated American.
- Corporations and the governments are using AI to spy on us and control what we see and sometimes what we can do.
- To improve the effectiveness of advertising and political persuasion without anyone understanding what it is doing.
Search: AI Chat in Operation
Here's where I find this technology to be useful. I setup some tests to show you what it does well and where it fails. I posted the same questions in You.com's chat and OpenAI's ChatGPT so I could compare answers.
I asked: Who is G. K. Chesterton?
Youchat: Gave me a clean one paragraph answer. If I'd asked for 3 paragraphs, the answer would be more complete. It was a good answer.
ChatGPT: A slightly better answer.
I asked:Describe his philosophy. Notice that I didn't need to say whose philosophy. You can interact and follow-up on discussions just like with people.
Youchat: Gave me a nice 1 paragraph answer. When I asked it to expand on that and give me 3 paragraphs, it gave me a larger paragraph which, while sounding good, wouldn't help someone understand. When I asked for more depth, it answered that it was still learning and couldn't explain more.
ChatGPT: Got a much more detailed 3 paragraph answer. OpenAI trained ChatGPT on a larger data set.
I asked: How did he defend Original Sin?
Both systems gave me good answers.
Both chatbots wrote their answers well and appeared to be real experts.
Now I decided to push them beyond what I thought they could do, so I asked:Did Vivekananda accept the Samkhya idea of multiple purushas?
This is a rather esoteric Indian philosophy question. Youchat's answer was entirely wrong. But it wrote in an authoritative manner that was believable. We need to be careful with these bots, and with every other source of information as well. There was an online lecture Swami Vivekananda gave on Samkhya and perhaps Youchat drew the false conclusion he was a Samkhyist. When using these bots, remember they can be wrong. ChatGPT gave me 4 paragraphs about half of which were incorrect.
I followed that up with another esoteric question: Explain the difference between Advaita and Samkhya. Youchat told me it was still learning and that was beyond its ability. ChatGPT gave me an excellent 6 paragraph answer.
I followed those up with these questions they both answered well.
- What are three types of windows I can install in my home?
- Are some windows triple paned?
Both got these right and explained the three types of windows and the functioning of triple pane windows.
Finally, I asked them both for two recipes for oatmeal cottage cheese pancakes and got what appear to be good answers. However YouChat only gave me one recipe!
I said, That was only one recipe.
YouChat responded: Sorry about that! Here is another recipe for oatmeal cottage cheese pancakes
In the new recipe, it replaced the maple syrup from the first recipe with 1/2 teaspoon of vanilla extract and 2 tablespoons of honey.
Both bots suggested using fruit in the mix and making sure the rolled oats were ground well before making the pancakes. I hadn't done that.
Conclusion: As long as you don't ask questions that are too esoteric, these bots give excellent answers and an excellent summary of important points.
The Problems With Chat Search
On the page with their bot, You.com says, "This product is in beta and its accuracy may be limited. You.com is not liable for content generated. "
Of course incorrect information is a problem with the life in general, not just the Internet, bots and government. Much of our news coverage in the last few years reported incorrect "facts". One of my favorite examples is a 1920 article by the NY Times called "A Severe Strain on Credulity" ridiculing the notion that rockets could go to the moon. To imagine that once the rocket left our air, they said, it would have nothing to push against and so couldn't accelerate or change direction. "To claim that it would be, is to deny a fundamental law of dynamics...". They did retract this in 2001, 32 years after Neil Armstrong walked on the moon.
In general, I feel pretty comfortable with information which is widely known and discussed, just not by me. I'll be getting a consensus Internet opinion. But when things get fairly esoteric or you ask about things few people are interested in, it might give false results, just like a search could.
How to Use ChatGPT and YouChat
For ChatGPT you need an Openai account. Once you have the account, you can log in and use their ChatGPT for free. For now, YouChat is simply part of the You.com search engine. You.com search has become my favorite search engine. It is privacy centered like DuckDuckGo but is prettier, and has a neat writing and chatbot. I expect to do an article on it next month.
More Information
- A negative and excellent 20 minute Ted Talk about the dystopian uses of AI.
- A positive glowing video: This is epoch times fantasy about how wonderful it will be and already is.
- You.com's YouTube channel. Lots of good information here about using You.com and especially their chatbot.
Date: January 2023
This article is licensed under a Creative Commons Attribution-NoDerivs 3.0 Unported License.