AI is the bastard child of predictive text.

Next-Right-Word of AI

I am a pretty good typist. When I have access to a nice keyboard I can fly. And predictive text is not my friend. It gets in my way. In iMessage (when I’m using the computer, as I am now) predictive text messes me up. Actually slows me down. I think it might be great for slow typists. It is okay on my phone, but here’s my beef with the entire generative text idea.

>> Listen to the NotebookLM Deep Dive Podcast of this article<<

Dear AI (predictive text too), You have no idea what my *human* next right word is going to be. Don’t try and read my mind. If I ask you to write something for you, I will give you the prompt. Otherwise, chill out.

Here’s the kind of *fk up* that happens when I write on my nice keyboard with predictive text on.

TEXTING: “Hey, I’m ready to leave for the movie.”

PREDICTIVE HELP: “Hey… how are you (delete delete delete). 2nd attempt, “Hey, I a… always (delete) am r… really (delete) 3rd attempt, “Hey, I am ready to l… earn (delete) le… let (delete) leave for the m… mamovie.” (delete the inserted predictive error, from the front of the word I wanted.)

This is a micro-example of how predictive text doesn’t get me. It’s the same with AI. It cannot read your mind. It has no mind or *intelligence* of its own. It has a massive LLM (large language model) but no emotional or human comprehension of the words. AI in predictive mode is math and letters. As fast and powerful as AI is, it does not COMPREHEND any of the words. Not one of them.

Where AI Loses

AI is a massively powerful tool. The entire planet is waking up to a new dawn. Fortunately, the SkyNet future for AI is science fiction. AI has as much *general* intelligence as Apple’s predictive text model on their iMessaging app. As a next-right-word process, predictive text can go in any direction you choose when you’re working with a platform like ChatGPT or Gemini. If you can think of a prompt, there is probably an AI that is built to work on just that idea. Workloads, they call it. Whatever workload you have in mind, there is probably already a GPT created to help you. Here’s one simple example I used in a webinar yesterday about Practical AI Uses Today.

These are GPTs that have been created to do very specific *machine-capable* tasks. With the exception of Humanizer Pro which is like an addict trying to show you how addiction is good for you, most of these are worthwhile explorations. You can find the GPT Market Place only after you’ve subscribed to Openai’s membership. Once you have subscribed you can build your own GPTs. It’s not too technical to get started. A bit more technical to tune or refine your GPT. A lot more technical to build one of the laser-guided GPTs above. But, AI models and GPTs are fun. Give it a go.

Where AI Wins

AI is powerful at generative content for (articles, software code, advertisements, resume optimizations, cover letters written to specific job descriptions, and most of what you can imagine asking it to generate. The new video tool from Openai, Sora, has shocked the world in the same way ChatGPT did more than a year ago. My Sora Example Video.

The GPTs written about how to make better prompts are a bit like Alice in Wonderland. You can have ChatGPT help you write better prompts.

Where AI Becomes a Threat

If AI wakes up and is unhappy with the human race we are in trouble. Terminator was written from this bad inflection. SkyNet was a satellite system meant to protect the Earth. When the AI takes over, it becomes a weapon of extinction.

Artificial General Intelligence is the idea of AI gaining some sort of self-awareness.

Today, AI may or may not be working to undermine our human existence. There are bad actors trying to influence AI. Nightshade, is an app that feeds bad information and bad prompts/training into AI. As we watch, ChatGPT 3.5 (the publicly available GPT) is getting dumber. The bad actors are training 3.5 to make the wrong predictions. It is having the predictable negative effect.

What Artificial General Intelligence Means

My favorite example of the limits of AI is Rachel in Blade Runner. Beautiful, seemingly self-aware. And a robot. Today, the AI we know is nowhere near Rachel. And even if we add quantum computing into the mix, a 1,000 X more powerful ChatGPT is going to be faster, better, and no closer to AGI.

What is needed is some human element. Today, that is: humans. The idea that AI will suddenly develop self-awareness and something resembling creativity is not possible without human intervention. I added HUG as an alternative to RAG.

With humans on-board AI is as amazing as the business world is making it up to be. In this exchange with Claude.ai, I experienced an emotional moment after the AI responded to the equations of letters that equalled, “my mother died.”

My Emotional *ai* Companion: More Conversations with Claude

I think from my Conversations with Claude, I saw the magic of predictive text. I saw a potential use as psychological chatbots that might point to a therapeutic solution for mental anguish. AI will not become good therapists. Even with all the knowledge of the largest LLM, AI cannot understand the word depression, death, or loss.

The Path Forward with AI

I’m not as concerned about the AI takeover. But it’s obvious AI can be used in cyberattacks, battle robots, and predictive battlefield logistics. AI is already weaponized. AI is not even close to becoming aware. I’m happy to update my own human awareness if you can point me to a “creative” or “human” response from an AI.

I am working in AI with my business: Iterativ.ai. I have a lot of words that I have written about AI from my own creative and human understanding. I’m even planning a conversational podcast Tecquil.ai about AI with AI engineers, artists, and other humans.

Let’s keep our minds open. And let’s push ai for good to the limits.

image prompt for dall-e – variation v 3 : photograph surrealistic bird’s eye view of a single robot woman, in a black leather dress in a whirlpool of green waves and sharks beneath the surface, put her face in the foreground of the image, she is wearing bright red sunglasses

John McElhenney — LinkedIn

Check out the new generative video tool from Openai – Sora – in this breathtaking video. The Ghost in You – Buzzie