#NotebookLM is crazy

In the evaluation of Google’s new #NotebookLM, I pointed it at the ColhounTech company website. You can get the AI to generate a podcast style 2 person discussion of any source, document, pdf, or website. I have to be honest – it blew me away.

Firstly, the voice tone and cadence quality. I could not tell that this was AI just from the conversation alone. It doesn’t sound like AI. At times I even heard one of the speakers interrupt the other speaker, just like a real person would. The gaps where they, the ums and ahs, and pauses. That was all impressive

But the structure and format was even more impressive. The “podcast” has managed to extract all the core features and points I was trying to make with the website. Now there is a point of interest here.  The Website was AI generated by pointing another AI agent offered by Astra (ZipWP AI) at a short version of my Business Plan. The postcast picked up on ClearVision(tm) Strategy and on the correct target market and made some excellent points on this market. It also picked out 3 items that impressed me.

Firstly, the ZipWP AI generated site had created some fake Customer Claims. I had set these to invisible, as I plan to replace these with real customer result, but the AI picked up on these even through they were set to Display:none. OK, that’s on me.

Secondly, I had added 3 glowing LinkedIn testimonials on the site, These are real and I asked each of them if they were happy to have the quotes on the site. The “podcast” host picked up on these, and while warned about self promotion testimonial, called out that they were consistent that this Michael might be the secret to ColhounTech – which of course is true.

Thirdly, it make a very interesting point. If “this michael character” is the secret sauce, then what is really important is that his extensive experience can be replicated across the organisation. This is quite deep for me, because that’s actually a core part of the business plan that was NOT in the generated site, our ambition and skillset in transferring knowledge. It was a great call because in some way it predicted the future.

Which brings me to the core point of this article. How good will AI LLMs become at predicting the future?

I guess only time will tell.

Here is the Podcast if you are interested in listening. It’s 4 minutes long.