Episode 34

When Should Businesses Trust AI to Work Without Humans?

Key Takeaways

  • AI is a useful sounding board: it can flag when you are making assumptions or getting too passionate in correspondence.
  • Use AI to draft page and blog structure, but do not publish AI slop: human writing is still the goal.
  • Whether to hand a process to AI depends on how damaging an error would be: a 95 percent success rate may be fine in one case and not in another.
  • Where one detail must be exactly right, such as a date, a human still has to check it.
  • Ask AI to build a curriculum and skills map to teach new hires or help you learn a new discipline.
  • For repeatable tasks like invoices, normal code and automation may do the job better than asking AI to write everything.
  • As AI-generated messages multiply, inboxes may end up talking to each other with no one reading what is said.

Main Topics

1HUMAN OVERSIGHT

When AI can run alone and when a human check is still needed.

2EVERYDAY USES

Farm management, weekend planning, web copy and calming a heated email.

3LEARNING WITH AI

Using AI to build curricula, skills maps and research briefs.

Featured Guests

JP

Uses Claude and NotebookLM for farm projects and for preparing book interviews.

Dr. John La Puma
MG

Uses AI to plan weekends and trains new hires to learn with it.

Michael Greenberg
BT

Uses AI for webpage and blog drafting and readability analysis.

Bill Treloar

Meet the Hosts

Elizabeth Gearhart
Elizabeth Gearhart, Ph.D.Chief Marketing Officer, Gearhart Law

Elizabeth Gearhart, Ph.D. is the host of Real AI Use Cases: Business Owners Roundtable, where she explores how businesses are actually putting AI to work. She is also CMO of Gearhart Law and co-host of the nationally syndicated Passage to Profit radio show and podcast.

Richard Gearhart
Richard Gearhart, Esq.Founding Partner, Gearhart Law

Richard Gearhart, Esq. is co-host of Real AI Use Cases: Business Owners Roundtable and founder of Gearhart Law, a full-service intellectual property law firm. A longtime podcaster, he hosts Intellectual Property News and is the host of the nationally syndicated Passage to Profit radio show and podcast.

Elizabeth and Richard talk about using AI as a sounding board, planning a software upgrade and where all the email will end up.

Episode Summary

Frequently Asked Questions

When can businesses trust AI to work without a human?
Michael Greenberg says it depends on how damaging an error is. If you can accept about a 95 percent success rate you can often hand a process to AI, but if the last 1 percent is mission critical, a human should check it.
Should you publish AI-written content on your website?
Bill Treloar says AI is an excellent start for brainstorming and structure, but you do not want to publish AI slop. You want human writing.
How can AI help train new employees?
Michael Greenberg says his team asks AI to set up a curriculum of videos, books and articles and to build a skills map toward a goal, which new hires then use to learn a new discipline.
Can AI act as a sounding board for tough decisions?
Richard Gearhart says AI helped him tone down a too-passionate message by pointing out assumptions he was making, giving a more objective perspective than a colleague might.
Is AI always the right tool for repeatable work?
The guests note that for something like invoices, regular code and automation that pulls data from your accounting software may be more reliable than asking AI to generate everything.
What concerns do the hosts raise about AI and email?
Richard Gearhart wonders whether inboxes will end up communicating with each other with no humans following what is said as AI sends more and more messages.

Episode Transcript

View Full Transcript

Now it is time for Real AI Use Cases: Business Owners Roundtable. I know we have been talking about how it is not great to be on screens all day, but we are all using AI. Dr. John La Puma, what is one way, your favorite way or your most recent way, that you have used AI? Just one for now, and then we will go around the table and do more.

I am an enthusiastic user of tech. I use Claude Cowork for all sorts of things, including preparing for our interview. What I like to do that is less traditional is put a lot of my farm inputs into Claude Cowork. I have a separate project for the farm where I am testing some of my observations about our trees and bugs and trying to think of ways to complement them and fight them with non-synthetic, natural methods. So I use AI to help manage my farm.

Wow, that is great. Michael Greenberg with 3rdbrain.co, we know you live and breathe AI, but what is one way you use it for yourself that you really like?

In my personal life, I like to use AI to plan my weekends. Especially during the summer, I like to drive out to farms and visit, and buy produce or meat from various vendors. The AI can plan not only where we are going and whether we can hit two or three places in a day, but also where to stop for lunch or breakfast on the way. That basic travel planning saves me a ton of time.

Excellent. Bill Treloar with rankmagic.com, what is one way you are using AI?

I use AI a lot for webpage and blog writing. I will give it some parameters and it will tell me, do not forget to mention this and talk about that, and here is a suggested structure. The trick is that you do not want to be publishing AI slop on a website. You want human writing, but it is a really excellent start for brainstorming and giving you ideas for how to write a page.

Excellent. Richard Gearhart with Gearhart Law, what is one way you are using AI?

Most recently I have been using it as a sounding board for bouncing legal ideas off of. I have been doing that for a while, and this last week it has been especially useful. There were times when I was getting a little too passionate in my correspondence, and AI said, you might need to throttle this back a little bit, because you are making some assumptions that may not be correct. I thought about it and toned it down. It is good to talk with colleagues, but having a more objective perspective sometimes helps. We all have our opinions and agendas, and I think AI does too, but it is probably not biased quite in the way some of my other sounding boards would be. So it has been very helpful.

And me, Elizabeth Gearhart with Gear Media Studios. We are doing a software transition, an upgrade. Last time we did it there were a lot of issues, and it all comes down to the people more than the software. We did not have the right people in the right places. I described all of our issues to Claude and to Perplexity and asked each for plans to help this not happen again. They both came up with pretty good ideas that we are actually implementing with this upgrade to try to prevent a big snafu, an upgrade disaster. I brought it to Richard, and we are not doing everything, because of course they tell you a million things. Oh, and do you want to find out more about this, go down this rabbit hole, go down that one. But we got the gist. My experience is still that you have to ask the right questions, because every once in a while you will say, what about this, and it will say, oh yeah, and change its output. So I think human beings are still required here on this earth despite AI.

Does anybody else have anything fun to say about AI, or any fears? The reason I am doing this segment is to help people start using it if they have not thought of great ways. Michael, you were going to say something?

One thing we try to train every new hire to do is use AI to teach them things. They ask it to set up a curriculum with recommendations of different YouTube videos, books and blog articles they can read on a subject, and to build a skills map, the individual skills that lead to the bigger skill you want to build. Then they use that as a project to help learn a new discipline or improve themselves.

Michael, do you think AI is always going to be under the thumb of human beings, or do you think it will take over the universe someday?

I am not sure it will take over the universe, but we have a lot of clients, and I have seen quite a bit myself of it doing plenty without a human involved.

I described a process we have at the law firm and asked whether an AI agent could do it. It went through the whole thing and said, you need to check here, here and here. I thought, if we have humans do it with those checks in place, what is the real difference? I do not know that in that case it would have made things more productive.

I think it comes down to how damaging an error is. Inside a process right now, a lot of times you can get to 98 or 99 percent. If that last 1 percent is mission critical, you need a human to check it. But in another case, if you are okay with a 95 percent success rate, you can often pass the whole thing off to AI.

What was nice is that it said, this is one thing you have to be absolutely sure of, like a date, so you have to have a human check it. You cannot leave it to AI. That is what Perplexity told me.

That comes down to data again. If you are sending out an invoice and having AI generate it, you might not want it to write the entire invoice. You might want it to run some code that pulls the invoice information out of your QuickBooks or other system, compiles it, writes a note and sends it. At that point it is not quite AI anymore. It is more normal code and automation that accomplishes the same function.

I want to talk about that when we get to your segment, because I have strong opinions on that. Dr. La Puma, are there other ways you are using it that are fun or that you like?

I use it before pretty much every podcast I do about my book, Indoor Epidemic. I have probably done 30 podcasts, so I have 30 different notes sections. I use it to think through helpful replies about the types of research that are emerging. Sometimes I use it as a research agent, particularly NotebookLM, where I give it all the resources and references, and that has gotten even better since you can include not just text but video and other kinds of assets. I am also finding the Google AI agent for Gmail very useful. They have a beta product called CC that looks at your email, reminds you of what there is to do and sends you a daily update. Sometimes I have to put four or five days into Claude and ask what I missed, so it can be overwhelming. But I think it is an evolving, smart tool, and people who learn to use AI well are going to keep their jobs and create new ones.

What about you, Bill? Any other ways you are using it that you like?

One of the things I have used it for is readability analysis on client web pages. It can say things like, your sentences are too long, or your tone is off-putting in this or that way, or you are using too many 11th-grade words.

I can see how that would be really useful. I used it this morning. I was mad about something, and I told it the situation and said, I want to send this person an email but I am really mad right now, can you write the email for me? It wrote a very fact-based email that I felt comfortable sending. What about you?

I am listening to all of this and thinking about the volume of email and communications we get every day, and using AI to address those communications and send more communications. I wonder where all this is going to lead, to the point where we have our inboxes communicating with each other and we do not even know what is being said.

That is a real possibility. I think that time is close. Thank you very much. This has been Real AI Use Cases: Business Owners Roundtable.

Keep exploring the show

Hear how more business owners are putting AI to work in their day-to-day.