2024 January 23 Can AI write this podcast?

Jan 23, 2024
 

 Hi, this is Jim Cranston from 7EveryMinute and 7EveryMinute.com, the podcast and website about reimagining your life. Thanks for joining me tonight to talk about AI. Is it really smart? Let's get started. If you like what you hear tonight, please leave a like, tell your friends, send me a message.

 

Tonight, I'm going to talk a little bit behind the scenes and the creation and production of this weekly broadcast. There are a lot of different software tools that we use, but only two of them really use AI and not to a large degree. But the reasons why I don't really use AI may surprise you a little bit.

 

It's certainly not because I don't see stories and get ads essentially every day telling me I should let AI write my podcast and give me ideas and save myself from working all the time. But let's remember what AI is all about at this stage. It absolutely does not think. It's clever, what's called an inference engine. 

 

Our brains are also inference engines, meaning that we come up with new ideas by logic and reasoning. We actually go way beyond that in forming our own intellectual associations. AI in general does not go much beyond the basics. While it's good at coming up with possibilities by expanding past knowledge, it isn't so good at coming up with totally new theories based upon that previous knowledge in a non linear or creative way.

 

This all sounds a little boring. What's that got to do with the podcast? Well, that inherent bias of just taking things and trying to extend what happened before leads to some other interesting biases. It's important to remember that the core input is lots of common knowledge from all over the internet and elsewhere in books. Most of it's not creative thinking in the sense that the AI engine isn't capturing the creativity. It's just looking at the result of that creativity. That’s a big difference. This leads to what I describe as AI's excellent ability to be mediocre. It's very good at appearing knowledgeable, but it only appears knowledgeable to someone who's not really knowledgeable in a particular subject. Since it's trained on the aggregate of common knowledge, even though it can apply some degree of weighting, it also isn't really very good at insight. 

 

There's a saying in business that if you try to target everyone, you end up pleasing no one. You generally can't make a product that appeals to everyone. Ferrari and Lamborghini, the fancy car brands, don't target the happy homeowner looking for economical transportation, yet they have a dedicated, maniacal following in the targeted audience of exclusive high performance car buyers.

 

Actually, they are now the largest electric car manufacturer in the world, at least in certain metrics. There’s a Chinese electric car maker that just introduced a Lamborghini lookalike, called the C60. I'm sure it'll get a following and it'll probably sell a bunch, but it's not the same people buying that as would really buy a Lamborghini, because it's going to be relatively inexpensive. It's a totally different target audience. 

 

But back to the podcast tools, I have two tools that I can use to make short videos in my entire podcast. They both go through and they work, in a sense, but they go through and they look at the podcast and they take out little pieces and they make them into video shorts.

 

I also have a hidden assistant, behind the scenes, who transcribes each of those episodes, which both those tools can also do. It also helps in identifying what pieces of the podcast are interesting. Why do I have her do it? I could have the AI tools do it, but she is endlessly better at it than the AI tools.

 

We've tried for a number of years now. It’s not just in accuracy, and she's a lot better, not just in accuracy, but in maintaining my voice, how I speak, my methods of speaking, and the consistency of my messaging. When I talk live, sometimes things don't come out quite right. She's the magic behind the scenes that makes the transcript read like I meant it to, and not how I said it. She keeps my tone, and she also looks for slips and mistakes and takes care of all that at the same time.  AI cannot do any of that effectively.

 

It can do some of it in a rudimentary form, but not with the insight and knowledge a real person can.  Does that mean AI is useless? Not at all. It often provides a useful starting point, but taken by itself, it often sounds a bit hollow. There are almost 400,000 active podcasts, and each one is slightly different, even if the topics are the same.

 

Each podcaster has something special and unique, their own personality, their voice, their perspective.  That's why an AI generated podcast tends to sound like AI generated content. It's also why AI generated technical content needs extensive review, because it often has fundamental errors.

 

It sounds great on the surface, but often it's not practical. AI is not useless. It has some great applications, but its strengths are rules based. They're not really creativity based. They can be churned through to provide lots of results and then go through and look at the results and meet a certain criteria.  They're doing that now, actually. I just read a story on battery compounds. It's interesting, because it came up with a battery compound, and everybody's all a flutter about it.

 

But a very similar analog had been written up the week before in NASA Tech Briefs. It turns out the one in NASA Tech Briefs is working pretty well. The one that AI created is not behaving quite as predicted. It’s mediocre at best. They're trying to tweak it. So, even in something like that, it's still good at coming up with core ideas, but they're not finished products. At this point,  real people are still making the real breakthrough discoveries.

 

I think they will outperform AI for a long time to come because they have a different kind of insight and different kind of creativity than just taking a logical set of rules and trying to extend them into the future. Will that always be the case? Maybe, maybe not. But logical reasoning and creativity are very different things.

 

The Wall Street Journal,  the weekend edition, had a special section on AI.  They had AI write a business plan.  As I would expect, in the very end it was kind of mediocre. They gave it a pretty extensive prompt, but they said to get a really good plan, they probably would have required tweaking the prompt to such a detailed degree that it would have been just as easy to sit down and write the business plan itself. AI is a great starting point, and it looks official, but it isn't going to replace genuine creativity or insight anytime soon. 

 

But back again to the podcast: I've tried a few content generators, and the results, even with semi-detailed prompts, vary from marginally useful ideas to utter rubbish. Why do I even try them then? Because I want to stay current with technology, and if a tool is developed that truly saves me time, then it'd be of value to me because then I could provide more information to you. 

 

The good side is that we truly do care about what we write, and we do try to give a unique voice and perspective on things. That's very important, not just for a podcaster, but for a business, and it's even more important for a civilization and a community. Different perspectives create complementary viewpoints, even when they're in conflict, and that gives rise to more creative thinking and more robust and more lasting results. 

 

Really well-generated content is something that really captures your imagination. It’s something that really moves people forward as a group. Communication is to transfer knowledge and to build community. If something isn't focusing on that, then you really have to wonder - is it really beneficial at all? You can't do that creativity without putting your heart and your soul into the process, and AI has neither heart nor soul.  

 

So that's a quick peek into one aspect of how the content is created, and how AI plays a pretty small role, because it really isn't that effective. Be careful in general about being too impressed with AI, especially those that are about things you aren't really familiar with. They look really impressive, but it probably isn't impressing the real experts. The other message I’d like to leave you with is that I'd really encourage you, not just with AI, but with anything, to look beyond the headline.

 

If you read a long article, in the last third of the story, the truth really starts to come out after all the initial attention-grabbing headlines.  We've been fortunate to have been given this wonderful, trainable, curious, creative brain. It works best when we use it, and when we use it a lot, because it just keeps getting better and learning more and getting better. 

 

Skip the binge watching, and instead focus on your vision, and your goals to move your life ahead. That'll be far more entertaining and satisfying in the long run.  Your homework (always optional) is to look up a recent story, particularly if it's on AI or any other breakthrough discovery.  Find a long story about it, and then really read the story through to the end. Once you get there, see if the headline really matched what the story was talking about. Extra points are just for fun. You rewrite the headline. to better reflect reality. Try to rewrite the first paragraph of the story in your mind, just to see how amazing it is. I think you'll be surprised. Enjoy it. 

 

That's it for the evening. As always, please remember the people in Ukraine and the Middle East. If you can and you're able, please consider making a donation either through UKR7.com or to the World Central Kitchen at WCK.org.  They work in disaster areas to help bring food to people who need it.

 

We talk about this a lot, and it bears repeating. One of the best ways to help yourself is to help others, because it takes you away from looking inwards and encourages you to look outwards into life. It gives you a whole different perspective on your own life and your place in the world. 

 

As always, thank you for stopping by. If you found something interesting and useful, please pass it along. Please hit that like button. If not, please drop me a comment as to what you'd like to hear. Have a great week. Remember to live the life that you dream of, because that's the path to true contentment. Love and encouragement to everyone. See you next week on 7EveryMinute and 7EveryMinute.com. Thank you. 

Stay connected with news and updates!

Join our mailing list to receive the latest news and updates from our team.
Don't worry, your information will not be shared.

We hate SPAM. We will never sell your information, for any reason.