Enjoy this little review of the current state of AI. If you want to skip to the hilarious part, click here.
You may think of AI (Artificial Intelligence) as something that’s coming in the near future. In fact, it has been with us for more than 20 years.
For instance, somewhere around 2004, I was interested in maybe buying a Dremel tool, a fun device for little carving, polishing and repair projects.
For the next two weeks on many of the websites I visited, I was shown Dremel tools. (Did I end up buying one? I did.)
The basic idea with AI is that it can learn. It will try to adapt to what is wanted by end users. In recent years it has become so sophisticated that the very people who invented it don’t know how it works. Or, more specifically, they know the foundational architecture, but they are often unable to decipher how it presents what it does.
AI often gets it wrong. This is called ‘hallucinating.’ It may just be little things, like I was recently told by AI that three of the United States each have the longest coastline, one of which is Utah.
What a lot of people who manage and write AI applications don’t yet realize is it needs feedback to learn as quickly as possible. It can be as simple as a toggle. Slide it to yes when AI gives you a good response. Slide it to no if you know the information is incorrect, or doesn’t answer your question.
Going a step further, the designers could provide text areas in which you can comment on what AI presents. AI itself can parse what commentary you write, and improve itself accordingly.
You might worry that people would purposely mislead AI with feedback systems. I believe that’s a minor issue. By majority rule and some other algorithms, AI could weed out the trash. This isn’t much different than Wikipedia. Anyone can write information, but if it’s wrong, the majority of Wikipedia users will make corrections.
A very common worry with AI is that it will take all our jobs away. That’s especially true when you combine AI with robotics, which is not only inevitable, but already happening.
If you’re a writer, inventor, composer, artist, you might think you should run for the hills. How could you possible earn a living? What if you’re a manager, safety analyst, car mechanic, food service worker, surgeon, or even a movie actor? Won’t AI eventually replace your job as well?
You bet! But here’s why you don’t have to worry: AI plus robotics are exactly the right tools to bring us prosperity. Your job won’t be necessary to grow food, build cars, and make clothes. All that production will be handled for us. Prosperity in abundance! There’ll be no need for anyone to hold down a job. Instead, people will be permanently retired from the day they are born. They’ll devote their lives to your family, travel, sports, art, music, whatever interests them.
I suppose the far future will be quite different than what we know now. Children have to learn to read, write, add up numbers, and know how the world works. That won’t be necessary 50 years from now or maybe even sooner. AI can already talk. It can tell us anything we want to know. Soon, it will be able to compose and show us videos on every special interest. What would be the point of reading and all that learning? So, in time, we may become stupider that we once were, however we may utilize our intelligence in other ways. Maybe we’ll become more empathetic and generally more tuned into what might be loosely called ‘love.’
That brings us briefly to the opposite of love. That would be war. War is a crude tool to take prosperity from others. Well, if we can have everything we need, and it’s all free, there’d be no point in wars. Some wars are about ideology, but I believe AI will solve differences in doctrine as well.
Some of the more cynical among us understand that war is about profit. If you’re the CEO of a company that can provide food to the troops, armament, ammunition and so on, of course you will advocate war. With AI and robotics bringing us all amazing prosperity, that motivation for war also disappears.
Back to the present. I did mention that AI often gets it wrong.
I asked the well-known AI text composer, ChatGPT, to create a pickleball romance story, which you can see at Pickleball Romance and ChatGPT.
Then, I thought it would be nice to use an AI art generator, Dall-E2, to create some accompanying art work. First I asked, “Draw a picture of pickleball romance.” This is what I got:
As you can see, the art program isn’t great at spelling.
I thought the illustrations were too cartooney, so then I asked, “Draw a photorealistic picture of pickleball romance.” As you can see, I got more weirdness:
Want to see more? Check out AI Attempts Pickleball Again.
After that, naturally I had to ask AI itself to show me some examples of AI getting things wrong. Here’s what I got, filtered to eliminate the nonsensical responses:
Give me hilarious examples of AI hallucinations
ChatGPT:
Certainly! Here’s a list of humorous examples of AI hallucinations:
Chef AI on Overdrive: A recipe-generating AI suggests dishes like “banana peel stew” or “chocolate-covered sardines.”
Weather App Gone Rogue: A weather prediction AI starts forecasting absurd conditions like raining marshmallows or sunny spells of molten lava.
AI Historian: An AI analyzing historical events concludes that ancient battles were actually elaborate dance-offs.
Translation Mishaps: An AI translating between languages starts converting idioms literally, leading to phrases like “the cat is out of the bag” becoming “the feline has escaped the handbag.”
AI Traffic Controller: A traffic management AI directs cars to “perform a barrel roll on the highway” for optimal traffic flow.
AI Comedian Gone Awry: A stand-up comedy AI bot delivers jokes with all the timing of a malfunctioning cuckoo clock, resulting in a stream of nonsensical punchlines.
AI Art Critic: An AI tasked with analyzing art becomes obsessed with abstract squiggles, praising them as avant-garde masterpieces while dismissing Rembrandt as “so last century.”
AI Personal Shopper: A shopping AI recommends bizarre outfits like a neon tutu paired with scuba flippers.
AI Life Coach: A motivational AI starts encouraging users to “embrace their inner sloth.”
AI Musician: An AI creating music begins composing symphonies that blend opera with heavy metal and dubstep, resulting in compositions that defy all musical conventions.
AI Fitness Guru: A fitness AI suggests workouts involving synchronized swimming in a bathtub.
AI Psychic: A fortune-telling AI predicts world peace within the next 24 hours.
AI Relationship Counselor: An AI giving relationship advice recommends couples bond over synchronized knitting.
AI Gardener: A gardening AI suggests planting crops in the shape of QR codes.
AI Linguist: A language-learning AI teaches users to speak in ancient Egyptian hieroglyphs or Shakespearean English, believing it’s enhancing cultural appreciation.
AI Detective: A crime-solving AI starts accusing household appliances of plotting against their owners.
AI Financial Advisor: A financial AI recommends bartering with sea shells and moon rocks.
AI Astronomer: An astronomy AI discovers new constellations shaped like famous celebrities.
AI Inventor: An AI invents gadgets like self-dancing shoes or a universal translator for communicating with houseplants.
AI Archaeologist: An AI studying ancient artifacts determines that Stonehenge was a giant sundial for dinosaurs.
AI Fashion Designer: A fashion AI suggests accessorizing with live miniature giraffes.
AI Environmentalist: An environmental AI suggests combating climate change with clouds made of cotton candy.
AI Pilot: A flight navigation AI recommends detouring through wormholes for faster travel or landing on the moon for a scenic layover.
AI Sports Commentator: A sports AI announces events like extreme knitting championships or underwater ice skating competitions, convinced they’re the next big sports craze.
These examples illustrate humorous scenarios where AI misinterprets its tasks, resulting in unexpected and often hilarious outcomes.
Note that the sentence above was also written by AI.
