The Real AI Question for Small Businesses

It’s been a rough couple of weeks for AI. (Not that it ‘feels’ anything. lol) Elon Musk lost his court case against OpenAI for going for-profit from its original founding as a nonprofit. The Pontiff released his encyclical filled with (valid) warnings about how AI can/should be used (and regulated). The White House blows hot and cold about regulating AI. And Meta fired 8,000 workers amid growing claims that AI-driven efficiency is reshaping staffing needs.

It’s a lot to take in.

For some, the onslaught is discouraging and triggers a reaction to bury your head in the sand.

For others, the tsunami of news and developments generates anger.

For me, it’s a lot of mixed feelings. But the hard facts are out there: AI is getting bigger and more predominant daily. From where I sit, anger, frustration and head-in-sand burying aren’t an option. Soooo….

The best way forward is learning how to use the tools wisely.

That doesn’t mean everyone needs to become an expert in using AI tools. But these are truly tools where the old adage “garbage in, garbage out” is true. And then some.

When I ask friends and clients how (or if) they’re using AI, I hear everything from expert-level power use to flat-out “I won’t go near it!” objections. Most replies land in the mid-range of asking AI questions or using it to write or craft imagery. Few have considered how it can automate their workflows or conduct deep research…which is quite understandable.

AI is complicated. Without careful prompting, it usually demonstrates heavy confirmation bias. It makes big mistakes, hallucinates and can be riddled with errors in its output.

Yesterday I was prompting to craft a photo-realistic, geographically accurate S.F. Bay image from MAI-Image-2, and look at this messed up mashup it produced:

AI = Toddlers and Teenagers!

AI the Tool, Not an Independent Manager

AI requires a lot of adult supervision. (It took careful prompting and multiple revisions to get AI to properly arrange a bay image with key elements all in the correct geographic location.)

Though more and more are dipping their toes into the water to give AI a try, they’re getting mixed results and wondering if it’s worth the hype. Will they save a few hours in productivity? Twenty? Hundreds?

For small businesses, success comes from being more competitive than similar businesses. The question is no longer whether AI is perfect. It’s whether you can afford not to use it wisely.

AI Capabilities

An AI consultant, Paul Conyngham (shown at top), built a custom mRNA cancer vaccine for his dog Rosie by chaining ChatGPT, Grok, DeepMind’s AlphaFold, and UNSW’s genomics lab to turn 350 GB of tumor data into a treatment with real results. His dog had been given months to live. Conyngham’s work resulted in one tumor shrinking by half after Rosie’s first injection. Today, he’s finalizing her 2nd vaccine and has saved her life.

An extreme example? Yes.

But it’s no less valid.

We’ve all undoubtedly heard horror stories about AI as well, from ChatGPT discussions that have sadly led to deaths and legal cases charging AI platforms with copyright infringement. Each user must decide for themselves if and when they’ll use AI.

I’ve included a few links below as resources to recent articles on AI.

RESOURCES

Pope Leo urges world to ‘slow down’ on AI in fervent first manifesto [Reuters, paywall]

Here’s The Guardian’s analysis of the Pope’s encyclical.

Artificial Intelligence and Human Intelligence [The Vatican’s Encyclical]

Is AI dulling our minds? [Harvard Gazette]

Anthropic Constitutional AI overview

Anthropic Responsible Scaling Policy

Choosing to Stay Human…means choosing when and how to use AI. [Ethan Mollick, Substack]

Artificial intelligence yields new antibiotic [MIT News Office]

Doctors, This Is Why Our Patients Are Using ChatGPT [NY Times, possible paywall]