A Realistic Look at AI

In a networking meeting this morning, the presenter (a marketing agency owner) discussed AI, how quickly it’s grown, and what implications are for businesses who use a lot of search engine optimization for their marketing. 

Everyone wants to talk about AI – I get it, it’s the sexiest new thing! As a productivity expert, I often have people asking me if it will help them in their business. 

AI may or may not be beneficial for your business and productivity. If you’ve already laid down the fundamentals of productivity, AI may be able to help you on the margins. If you’re not already maximizing your brain power, you’ll hit a ceiling beyond which AI probably can’t help.

I’ve found it useful when analyzing, say, all the 1 star reviews for productivity books on Amazon. It can help a lot when you’re editing, and can also suggest alternative names when you’re playing around with an idea. I never use it for drafting posts or articles (and you probably shouldn’t either!)

The current state of AI for professional service businesses

AI as you know is short for “artificial intelligence”, yet it is not truly intelligence, artificial or otherwise. Most business owners are using AI in the form of large language models (LLMs) such as ChatGPT, Claude, CoPilot, Gemini, etc. All these models have been built on stolen data: copyright and other IP (intellectual property) protections ignored for the purposes of “training” the models on very large datasets.

There are other ways to use non-LLM AI. You can use different models for images, human voices, and more; some are more useful than others, like AIs that are “learning” to read X-rays and other medical uses. But I think most business owners are looking at LLMs so we’ll focus on those here.

You might be wondering (as I was) what the future for organic search results is now. Every time I search something in Google, I get the AI summary first (I need to go in and shut it off.) Below the AI are the paid ads, of course, and then organic results. The AI summary typically fills my screen – it’s like the top of the front page of a newspaper, “above the fold”. How many people scroll further down?

The marketing expert recommended very long-tail keywords, since those don’t often get AI overviews. She also said that results for local businesses often don’t get AI overviews either. I’m left wondering how long those keywords would have to be, and how clients or prospects would know to search for them. Maybe there are some edge cases where that would work, but I don’t think it’s useful for most businesses. And since many of us work virtually, the local trick won’t help either.

So it appears to me that organic search is over unless you have popular long-tail keywords and/or a local business. Why bother going through all the PITA of search engine optimization (SEO) when there’s no chance your business will be on the 1st page of results? I’d be leery of paid ads too, given this context. 

Not only does AI dominate the search results, it seems to be dominating LinkedIn as well. Posts talking about AI (including this one!), offering AI help, giving away prompts, and so forth. But this ain’t my first rodeo – the situation is very familiar.

Remember the dotcom boom? And as night follows day, the dotcom crash followed the boom. A new sales channel, no matter how sexy, doesn’t mean the economic laws of supply and demand suddenly vanish. 

During the dot-com era, I was in finance. I spent my time telling clients to diversify out of tech and to avoid investing in $hit like 1800sockpuppets.com. Even though at the time I was relatively young (mid to late twenties), I knew perfectly well that the high valuations would crash. 

As they did.

Before the Great Recession, I was thinking about buying a home. But I couldn’t bring myself to pull the trigger because I knew house prices were too high and were due to come tumbling down.

As they did.

This time I’m not making any predictions about the finances. Most of the LLM models are backed by existing well-capitalized firms. I’m not expecting the companies to fall apart or anything like that.

It’s the bandwagon effect that’s the same. Many people using AI models aren’t aware that they’re using stolen data, and some believe that the machines are thinking. Don’t worry, they’re not, even though I’ve seen plenty of media that seems to indicate otherwise.

Tell me if you’ve heard this before: if you don’t jump on the bandwagon, you’ll be left behind and all your peers will zip ahead of you. Honestly, I think this hits older people like me harder, because we’re worried about being left behind by tech. However, this is pretty much the exact same argument that was used in the ‘00s regarding mortgages, and you see where that led us.

This particular bandwagon, unlike previous ones, requires huge amounts of water, electricity, and land for the machines to process all the data. Human garbage fires can remove all references to climate change as much as they want, but that doesn’t actually make the problem go away. 

Problems with AI that could interfere with improving your productivity

My recommendation to all business owners is to think through how you might be able to use it, the results you expect, and a policy for your company determining how and when people use AI.

I think you do need to disclose how and when you use it. 

To me, some AI slop is obvious. There’s a radio ad where the woman’s voice is just slightly too robotic to be real. I don’t even want to think about all the horrific images that are showing up on Facebook that are simply hideous.

It’s sometimes a little harder to tell with written posts and articles, except that the writing has no personality. It’s slopped down to the lowest common denominator. One big tell, but only for those who pay attention, is whether stats and citations are “hallucinated” by the model. 

This is an issue for everyone, but especially for those in the legal field. You don’t want to get in front of the judge to find out the case you cited in your documents doesn’t actually exist. (Unless that’s your kink; FSM knows I don’t want to kinkshame anyone.)

I think you also have to be careful which model you use. There’s a “system prompt” that companies give their LLMs for its overall quality; for example, ChatGPT is known to be overly sycophantic to the user.

That’s not as alarming as the system prompt for the AI that’s used on the flaming dumpster fire formerly known as Twitter. Once it prompted mention of (false) scenarios of white South Africans having their property stolen in every post. Once it called itself MechaHitler. You probably don’t want any of that getting into your articles and social media posts.

Using an LLM to write or draft anything first is probably a bad idea (I’ll get into the specifics of that a bit later.) But using it for analysis, editing, etc. could be very helpful. Likewise, brainstorming ideas to get yourself started can often be a good use of AI.

You might find that once you’ve added in the time you need to get a good prompt and the time you need to check any stats or documents that it cites, plus the need to rewrite into your own voice, AI might not save you all that much time in the end. But each person’s mileage will vary.

The future of AI

Magic 8-ball says….

Who knows? Had AI appeared say 5-10 years ago I might have been a little more optimistic about how AI could help us, especially business owners. But I have concerns.

Environmental impact

As noted earlier, the amount of power needed to run generative AI is massive and will likely make the climate change problem worse. I live in a state with frequent droughts and the amount of water necessary for even a tiny little question is ridiculous.

Further ensloppification

Social media platforms have, in the words of Cory Doctorow, become enshittified. Too many ads, too many bots, and the platforms don’t care. I’m calling my prediction for LLMs ensloppification.

LLMs were trained on existing material, but as more and more people use them, more AI slop will be created. In turn, the LLMs will include more and more AI slop in their “training” material. You don’t get better results by adding more slop, but that seems to be the path here.

Potentially increased rates of dementia

I don’t know if you saw it, but there was a study from MIT showing that people who used LLMs to write an essay couldn’t explain their arguments later, whereas people who wrote their own could recall theirs. 

Continuous learning is one of the things that seems to prevent or delay neurodegenerative conditions like dementia. If you’re constantly outsourcing your thinking to AI, then you’ll lose that capacity. If you’re no longer learning and thinking, then you’re more likely to end up in bad shape cognitively later in life.

Usage for evil

I was there in the early days of the internet. Remember the BBS (bulletin boards)? No pictures because we were all on dial-up internet, but we could make emojis out of punctuation marks like so: :-). Back then, we thought the internet would bring about more democracy, that everyone would have access to the information they needed, and so on.

Now we’ve got social media platform owners who praise genocidal maniacs like Hitler, bots from hostile countries that disseminate lies, people who envelop themselves in their bubbles so they don’t have to deal with everyone else. The loudest and nastiest get all the attention. 

We’ve already seen one case where a parody account (who even watermarked their video and said they used AI to create it) of a Congresswoman was used by a garbage human to attack said Congresswoman over what she “said” in the video. Even though she had never said anything like that before, because again, it was a parody. But the garbage human in question wanted to believe it.

And we know that hostile countries such as Russia create disinformation campaigns that are effective. Can you think of any way in which AI is going to make that situation better? I can’t.

Lack of critical thinking population-wide

The more people use software to think for them, the worse their skills will get. It’s bad enough already, with everyone in their own social media bubble. I thought that I was free of the bubble because I do know what other people think, but I also thought Kamala Harris was going to win big. Because in  my bubble, everyone supported her.

Thinking, problem-solving and the like: if you don’t use it, you lose it. It looks like there are plenty of people out there diving headfirst into the abyss.

How to use AI ethically for improving productivity and life

In an ideal world, AI doesn’t take over the creative work that humans have enjoyed for tens of thousands of years: art, poetry, etc. It takes over the crap stuff, like getting your car registration renewed or getting your defective fitness tracker fixed. It runs necessary processes for businesses in the background, so that business owners can fully concentrate on delivering for their customers.

Which gives everyone more time to rest, recharge, do things they enjoy, and spend more time with their families. 

Who am I kidding, we’re in late stage capitalism, life will get more and more miserable for most people. Unless you resist the pressure to work until you burn out.

I think, even if we don’t reach my ideal state, that we should lean into our humanity. Truly listening. Collaborating with each other. Getting to know clients, prospects, and colleagues on a deep level. Writing and making art that’s so personal it stands out among the AI slop.

I also believe every business owner needs to think about how AI might best be used in their business. You may not use it at all, and that might be OK! What works for you might not work for your friend who has a similar company, and that’s OK too. Instead of jumping on the bandwagon and hoping it gets you somewhere, look at your processes to see what can be automated and where AI might be really helpful to you. 

Recap (tl;dr)

We’ve been here before, so try to use AI thoughtfully and well. There are cases where you’re better off not using it, but there may be tasks or systems that would be faster and stronger with AI woven in.

I’m not an AI expert, but I am a productivity expert - in the form of sustainable a$$-kicking. Want to partner with someone to think through your AI use policies and where your systems might benefit? Click here to schedule your free consultation.

Photo by Emilipothèse on Unsplash.