Google NotebookLM and Our AI Toolkits

Welcome to another episode of AI goes to College.

I'm Craig Van Slyke, one of your co hosts, and I am joined by my friend and colleague, doctor Robert E.

Crossler.

And we're going to talk to you about a couple of interesting things.

We're going to start off by talking about Google's notebook LM.

If you have not heard of this or used it, you're in for some interesting insights.

And then we're going to share our personal AI toolkits.

What is it that we use, and why do we use each one?

So, let me start out by talking a little bit about what notebook LM is.

And so this is, we'll put a link in the show notes.

It's kind of a mini retrieval augmented generation system, which is a fancy way of saying you can upload documents, links to Google Docs, links to web pages and notebook will answer questions or create things based on that content that you uploaded.

So that may not sound like a big deal, but it is a huge deal.

You can upload up to 50 documents into notebook.

I don't think I've gone more than about six or eight, but then when it creates things, it creates things based on those documents.

Well, Rob, have you been using this at all?

Yeah, I've used it in my doctoral seminar.

It's been the primary place I've been using it, and it's been.

It's been pretty neat.

It does a really great job of taking what you upload to it and letting you ask questions of it.

We started in the doctoral seminar I was teaching, having the students do two different things.

One, they read all of the readings for the week and summarized it, did their own assessment of the articles.

Sure they did.

They mostly read all of them.

Yeah.

And then they started asking notebook questions of the papers that they had read and they'd done their own summaries of.

And what they told me is they really appreciated it because it gave them more confidence that the assessment that they did was correct.

So they ultimately felt when they came to class, that they were verifying for themselves that they knew they had done a good job in preparing for the class.

The following week, I had them upload the documents first and ask questions of the documents first and then read the papers.

And they actually found that less helpful because what that did for them was in some ways what they're going to assess from the papers.

It had already done it.

And so they've maybe found themselves a little less critical of the information it was giving them because they trusted it from the week prior.

So I want to drill down on that a little bit.

I think that's an absolutely critical difference, because as teachers, as professors, we want to encourage our students to use generative AI to supplement their learning, not replace their learning.

And if I could kind of summarize the two experiences.

In the first scenario, they had done the work of learning, and then it was just confirmation.

It was like they all went out and had a beer, had a cup of coffee, and started talking about the papers, and, oh, wait, you got this out of it.

Well, I got that out of it, and it was kind of simulating a little bit of that sort of a conversation on the other end of it.

The second scenario, they were just kind of letting notebook do the work.

And I think that's an absolutely critical difference, whether it's for our students or for our own work, even administratively.

I think if you're trying to upskill or something like that, you got to be careful in how you use generative AI.

Well, and then what made it very interesting was the next week, we took it a step further, and what we did was I had the students, after we'd done the week's readings, to take and upload the six papers and have it create the podcast.

So one of the things that notebook lM does is it'll create a podcast in NPR style with two people that sound better than me and Craig, and, oh, Craig disagrees, but.

And then to share it with the class, we could all listen to the podcast that each person created.

And one of my students ended up with an eight minute podcast.

The other one ended up with a 16 Minutes podcast, and the eight minute podcast excluded two of the six papers in how it talked about it.

So it didn't include everything.

The other student ultimately had created several podcasts to get to the point where, by changing some prompts and how it asked it to do, it got notebook LM to actually talk about the other two papers that were being excluded.

It made the podcast twice as long, and part of the reason was two of the papers were a little bit more tangential to the topic, so it was not a clear connection.

And so I found it was interesting, if you're relying on notebook LM to get it right, the one that was shorter was not incorrect, but it was not inclusive.

And so it took knowing what was in there, knowing something was missing, to know that you had a subset of what could have been and how it took extra knowledge and extra effort to get it to do all of the work.

Interesting.

Yeah.

And there's also, the non deterministic nature of these tools.

So it's not like a regular information system where you put the same input in ten times, you get the same output.

You can literally do a prompt, start a new prompt, put the exact same prompt in again, and you will get a different answer pretty much every time.

Often that's not a big deal, but every once in a while, that sort of thing matters.

And so we need to be a little bit careful about that.

By the way, for those of you who haven't tried this yet, the quality of the podcast, or I think they call it the audio overview, is shocking.

I would challenge anybody to tell the difference between two humans talking and what notebook produces without really paying attention.

It will even do things like have breath noises.

Every once in a while, it'll kind of tap its microphone or bump into its microphone, or pop a pee or something like that.

It's really scary how good it is.

But I can see a lot of pretty good uses for this, even on the administrative side of things.

You know, you've got some new policy that comes out, or a set of policies.

You can put those in, create a little audio overview, or they've got what they call their notebook guide.

If you click on that, it gives you some pre built prompts.

It's a briefing document, frequently asked questions, table of contents.

Oh, a timeline, which is kind of weird, and a study guide, which may or may not make any difference depending upon what you're trying to use it for.

But they're all pretty interesting.

So I use it in a little bit different way, and I know my students are using it now.

I have not prescribed to them how to use it, although that might be worth trying.

I use it to do my own preparation, so I do the same sort of thing.

Rob, you talked about put all the documents in, then I create the briefing document, that sort of thing.

But I ask it for the key takeaways.

I ask it for.

This is where the big payoff is for me.

Where are my students likely to get confused?

So, from my seminar, it's the very first one they have in their program.

And so a lot of things that won't be confusing to them later on, Orlando, might not be confusing.

Rob, to you or me, is gonna be really confusing for them.

And so sometimes we get blinded by that.

We're talking about a doctoral seminar here, but it could be the exact same thing in an HR training program you do over and over again in an undergraduate class.

You know, we just.

We know this stuff so well.

We don't understand where the confusions might come in.

And, you know, three quarters of what it comes up with are things I would have anticipated.

But that other 25% making up numbers is pretty golden for me as an instructor.

So it's a fantastic tool.

My biggest fear is that Google is going to take it away at some point, as they are wont to do.

But if you have not checked out notebook lm, you should.

And I think it's notebooklm dot google.com dot.

I think that's it.

That is it.

But we'll have a link in the show notes rob I'm even using this playing around with my kind of philosophy podcast, live well and flourish, which is available@livewellandflourish.com.

i took some episodes I did on self leadership and put the scripts in and then ask it to create some documents and to create a podcast based on that set of scripts.

So I'm kind of playing around with that as well.

So I think it's a fantastic tool, one that deserves a lot more attention than it's gotten, although in the last two or three weeks, there's been a lot of buzz about it.

I've also heard Craig to expect some more things coming from notebook LM, one of which is being able to create audiobooks instead of kind of the podcast sort of approach.

And also some of the issues, the limitations I mentioned of some of the papers, some of what you've uploaded being ignored as part of the audio overview, that they're going to make some tweaks, and that stuff is just going to get fixed behind the scenes.

So they're continually tweaking this, trying to make it better.

So I expect we're going to see something improving in it very, very soon.

And I hope they give us some settings for the audio overview.

Like, I could see if you've got four articles that are all kind of on the same topic, and then two outliers.

I might want a podcast just on the four main articles, though.

The other thing I forgot to mention that I have it due is my students have to write a synthesis paper each week, and it's a struggle sometimes because they want to summarize, although they're pretty good at synthesizing.

Well, I ask it to write the synthesis paper just to see what it'll come up with.

I give it the parameters for the assignment and just kind of see what it'll do.

It's usually pretty good.

All right, so again, if you haven't checked out notebook lM, you ought to.

It's it's really easy to use.

I mean, you don't have to do much prompt engineering to get the basics out of it.

So I would, for the fourth time, encourage you to check it out.

So that leads us into our second topic, our personal AI toolkits.

And we both use notebook LM, so that's one of them.

But what else do you use, Rob?

As I've mentioned in previous podcasts, I still am playing with copilot.

It's doing a lot for me that works really well.

I'm giving a keynote speech later this week on generative AI in the academic space in Africa, and I didn't quite know what that outline needed to look like.

And the copilot created write in Microsoft Word for me an outline that I go through, and it's like, yeah, that hits what I think the topics should be.

And now I'm filling in the blanks with some of my own research and work to be able to give that it's helpful in, again, communicating, writing a lot of emails and various things of that nature.

But one of the things that I found that it's not as great with, and we're playing with some analysis of some data and trying to have copilot take the data we'd given it and do some of this text and summarize and put things in different buckets and didn't like the results there.

Just felt like their surface level, it wasn't as good, and the iteration didn't seem to remember what it had done.

And so we went into chat GPT and did the same thing there and found it did better.

And so there are certain tasks where when I'm really wanting to iterate on data that I've given it, that I'm playing with, is it chat GPT?

Is it Claude?

And they both operate just a little bit differently, and I'm not sure which one is better than the other.

And I think it goes back to the I would be fine probably if I just picked one and went with it.

Copilot's funny because its superpower is that its embedded in the office apps.

Google's got something kind of similar with Google Docs, so it wouldn't surprise me if it's a little bit handcuffed versus just using chat GPT.

I actually had a breakthrough with Copilot, so I have not been a fan.

Okay, so you weren't a fan, but you're now a fan.

Did something change?

Well, I'm still not sure I'd go all the way to fandom, but I'm not utterly disgusted with it.

What I've wanted it to do is I want to be able to upload a document and have it create a draft of a PowerPoint.

I think we've talked about this before, and I've tried some various solutions, none of which were great, but I could never get it to work.

And then finally, this is a little tip that maybe everybody else already knew, but I didn't.

I created a blank presentation, stored it in the same onedrive folder as the document I wanted to do a slide deck for.

All of a sudden it worked.

And it worked fine.

It was okay.

I mean, I think it probably got me.

I was just playing around.

It wasn't something I actually had to give a talk on, but I think I probably got 50, 60% of the way there, which is pretty good.

I mean, I'll take that all day long.

The design was kind of crappy, but that's easy to fix.

I was okay with it.

I also had it create.

I asked it to explain how large language models work, and it was okay.

It was okay.

That was not based on a document, that was just based on whatever it thought the explanation should be.

And it was.

I didn't compare it to chat GPT.

My suspicion was it was more extensive than what chat GPT would have given me.

Plus it already had the slides, and.

The slides are formatted, aren't they?

I mean, have you seen that where it formats and makes them look nice?

You don't have to spend a lot of time on all that.

Yeah, that was nothing overly pleased with the format.

Like the text was way too small and the slides were way too wordy.

But that's, you know, that's easy enough to fix.

That's not that big a deal.

But what was interesting is I didn't use any of the slides, but it kind of gave me a little bit of a different view of how I could talk about large language models to.

I'm giving a talk to our college's doctoral students tomorrow to kind of, you know, here's a good way to explain it to them.

If you ask me, you've got $20 a month.

Do you want to put it into copilot or do you want to put it into chat GPT or Claude?

Copilot would.

It would not be a good day for copilot.

It would be way down on my list.

But I can see the value in it now more than I could before.

Yeah.

And where I've seen the most excitement about Copilot is when I talk to the administrative assistants in our college.

They are excited about some of the efficiencies that they think they'll probably be able to get with that because we're still in the evaluate it and make sure we're going to get it mode.

And I think we are, but the people who don't have it want it that spend a good portion of their day dealing with word and dealing with Excel and PowerPoints and those sorts of things.

Yeah.

And we have to keep in mind that we're still in early days.

This hasn't been out that long.

Matter of fact, I think they just released an update to copilot in the last couple of weeks, and so it's going to get better, and I'm sure that at some point in the not too distant future, it'll be pretty useful.

Anything else on copilot?

No.

All right, well, let's talk about chat GPT and Claude, because you said, if I heard you correctly, you're still not sure which one of those would be your go to.

Why is that?

They both have value and do things a little bit differently, and I don't know that one's better than the other.

They're just different, right?

Whether it's the the voice it uses, the details they pick up on and hang on to, they're both about the same speed, and so I'm not sure that one's better than the other.

And so most of the time I default to chat GPT because that was the first one I've used.

I haven't seen that Claude has pushed it out of the way.

In my experience using it, I find.

Claude a little bit better than chat GPT on a lot of things, but not enough where I've given up.

Chad GPT.

So let me tell you where I kind of use one over the other.

So generally, if I want to create something we talked about this before, I kind of like Claude because it will give you that artifact view on one half of the screen, and then you've got your chat on the other half.

OpenAI just released the same sort of thing for chat GPT although the name of they gave it is escaping me at the moment.

It has not worked as well for me, but it's okay.

It's okay.

So I still default to Claude for that.

I also think that Claude gives me some better answers sometimes, but chat GPT just added a new model, actually, two new models, zero one and then whatever, they're zero one mini or something like that.

I really should have checked all these names before we jumped online.

Zero one is more of a kind of chain of thought reasoning model where it will step you through how it's thinking about things.

And so one of the famous problems that generative AI has not been able to solve is how many r's there are in strawberry.

So that's almost a meme.

Could never get that right.

Well, zero one does.

So zero one is a lot better at certain types of tasks if it involves more complex reasoning and they've got some benchmark, another name that escapes me.

But most of the models were in the upper eighties, and I think zero one is now in the low to mid nineties on this, and so that's a pretty significant increase in quality.

So I think that might be a something that, for a lot of people, might tip the scales in favor of chat GPT right now.

But I would be shocked if Claude doesn't release something similar before too long, because that's the way these things go and it's good.

Yeah.

One thing I think that's going to drive people's use of these, Craig, is, as institutions do, wholesale adoptions of the different technologies.

So we're in the process, at my institution of looking at copilot, which I believe will be adopted, as well as the enterprise license for chat GPT.

And so once they're paying for chat GPT as an institution, to jump in and say, but, yeah, I need Claude instead.

I think it's going to be an interesting political conversation, of which different ones does the institution need to be getting into?

So I'll be interested to watch all this unfold, because I'm not sure what institutional licenses Claude has, and I haven't gone down that path because that's where we're not going as an institution.

It's also worth noting that chat GPT is more flexible, like, Claude is not going to create images for you.

There's another reason that I sometimes turn to chat GPT.

Its voice mode is amazing.

I mean, not just good, it's amazing.

Especially on the mobile app, you feel like you're talking to somebody.

Funny story with that, Craig.

I was playing with that the other day.

My wife was still in bed and I was sitting out in the living room having a conversation, and she gets up and she's like, who are you talking to?

And I was like, oh, I was playing with this new technology, and she literally thought I had someone else hanging out with me in the living room, having a conversation at 730 in the morning.

What are you wearing?

Right.

We didn't go there, thank goodness.

No, it's almost freaky.

I was just using it today because I didn't feel like typing a bunch of stuff.

And there's also, I've been thinking a lot about friction, how the technology gets in the way of the thinking, and sometimes even something as simple as typing can get in the way of the thinking.

And so to just sit back, you know, literally lean back in your chair and have a conversation is a lot more natural for some things.

I mean, other things it's not.

But for when you're, when you're trying to figure stuff out, like, I was trying to think through something, it just seemed a lot more natural to have a conversation, even though I knew, you know, it was this artificial entity, but it doesn't seem that way.

And that was on the Mac version, it's borderline freaky on the mobile version.

So, Craig, when you do that, does it give you a summary of the conversation later on, or do you have to remember, if you talk to it for half an hour about a thing as you process through, did you have to remember the conclusions you came to?

Well, it gives you a transcript, but that's actually a really interesting question, because one of the things I often do and did in this instance is I ask it to summarize the conversation.

So you found that helpful?

Did that give you a nice takeaway to have later on?

Yeah, it did.

It did.

And it gives you a transcript.

So if you go back into your chat history, I don't think you can replay the audio, but you can read all of the chat that you had, and so you can go back and look at it just like you could any other chat that might lead me to favor chat.

GPT over Claude.

But before we give the impression that those are the only two, Gemini is quite good at certain things, and we've talked about that before.

Gemini for tech support.

I had tractor problems or lawnmower problems this weekend, and I need a battery for one of my.

This is so sad.

I've got tractor, a garden tractor and a zero turn mower.

And so the garden tractor, the battery's dead and I need to get a new one.

Well, you know, I take a picture of the battery and say, what is this?

There are different sizes and I couldn't see where it said what size it was on the battery.

And it says, oh, it's a u one l.

Okay, well, that's what I'm going to order.

And was it right?

So it's great.

It was right?

Yeah, it was right.

So it's really good.

At that kind of thing.

We've talked about excel functions, really quick stuff, especially if you need it to surf the web.

No surprise, it seems a lot better than that, than chat GPT.

Even though chat GPT does have the ability to go out and search the web, it doesn't seem to work nearly as well as it does with Gemini.

I think most people would be perfectly fine with the free versions of all three of these tools.

Yeah, and that's what I hope Craig is, is the free version is going to stay usable.

My great fear is this turns into the world of all the streaming channels I pay for and, you know, a death by a thousand subscriptions.

Are we going to get hooked into using these different technologies and all of a sudden find ourselves shelling out, you know, $100 a month for all the different things because they all are better at one thing than another thing.

So definitely worth paying attention to of is the value add from adding one more really worthwhile if we get into a world of everything requires a subscription.

Well, I think you're right.

I think for most people right now, I would either get Poe poe.com comma, which is one of my favorites because of its flexibility.

And we talked about that in an earlier episode.

Poe gives you access to a bunch of different models, or just choose chapter Claude, play with the free versions of both.

Pick one if you want to start paying and use that one, because I don't think there, I mean, it's almost like, God, this could get me shot down here.

You know, the Ram versus a Chevy versus a Ford pickup truck.

You know, does it really matter?

Yeah.

One's better at this, another's better at that, but they're all going to haul hay for you and pull a trailer and do all that kind of stuff.

So just pick one.

But I would also use Gemini.

I have Gemini up on my screen much of the day because I turn to it for just really quick stuff.

Not so much helping me develop ideas, helping me develop content, but it's just questions, things that I used to go into Google and search well.

And that's what it sounds like.

I hear you saying, craig, is that you found Gemini has become your Google replacement in a lot of ways.

And those things that you were going to Google, you now go to Gemini.

And as you reflect on that, how long did it take you to change those habits or that inertia of always using one to move to that other?

It did not take that long.

Once I saw the kinds of things that Gemini was good at, I just started going to Gemini now.

You know, I wanted to check the score of last night's NFL game.

I just went to Google for that.

You know, why write a prompt?

I just type.

I literally type in NFL and the scores pop up.

And so there's still a lot of stuff that I use.

Google or Duckduckgo, since we're privacy people.

Duckduckgo for.

But if there's a question that involves anything technical, and I don't mean like it, I mean like, what battery do I need for my garden tractor?

I tend to go to Google or not to Google, but go to Gemini now.

And it did not take me long to make that transition.

It's just so much better at it than just a Google search.

You don't have to wade through all the garbage to get to what you need.

So what about more specialized tools?

So have you checked out perplexity AI?

I've not been to perplexity.

I've been looking at some related to research, research rabbit and some of those which are interesting, but I haven't played with them enough to say I sponsor it.

I say go use it.

But there's some really cool things out there when it comes to connecting papers together and doing some things from that perspective.

I've tried illicit, which I think has some potential, but it's still kludgy to me.

That's what I would encourage anyone to do as these tools come out and you're curious, is to play with them and see where they're useful, but to also, while you're doing so, ask the question of what am I missing?

What is it not doing for me?

And to be critical of it, not to just say, oh, wow, this just did my entire job for me.

But to say, did it do it right?

Do I trust it?

How accurate is it?

So I'm skeptical, but it's kind of interesting to see where technology's going with some of these tools that take the drudgery out of work.

Right.

That's really what it is, is when it's intellectual work and it's reducing the amount of effort it takes.

In some ways, if I can trust it, it's going to be a wonderful thing.

Yeah.

And that's the big problem, is can you trust it?

Right now, one of the big issues with those kinds of tools is that things behind a paywall may not get included, and that's a lot of our top journals.

Well, I've heard in some forums where people are saying when they find something behind a paywall, they go to the generative AI tools and ask it to summarize what's behind the paywall, and they're able to get access to things because it's behind a paywall.

So I see some interesting back and forths on whether or not our world of paywalls and how they work is going to keep working that way.

Given some of these things, I'll have to try that.

So perplexity.

And again, this is perplexity AI.

It's generative AI that cites its sources.

Basically, when you run a perplexity search, and it's got several different modes, if you go to aigostocolllege.com and go to the newsletter, I wrote a little article about it a few months ago, but it gives you little footnotes, and you just click on the footnote and it brings up that reference, and you can link out to the reference, which is kind of nice.

I find it really useful for getting a quick take on something.

So let's say you're interested in the ethics of generative AI.

You can go under perplexity and say, you know, what's the current research around the ethics of generative AI?

And it'll come up with a decent answer for you.

So it can kind of get you started if you look at it in terms of us as scholars.

But I think for a lot of more casual searches, it may be plenty.

And it has different modes.

It'll search the whole web.

It'll just search academic sources.

There are two or three different, I don't know, maybe, maybe six different modes that it's got something like that.

It's pretty good.

So that's another one.

If you haven't checked it out, check it out.

And the best way to describe it is it's generative AI that cites its sources.

And then there's a similar tool called consensus consensus app.

And like perplexity, consensus will kind of cite its sources, but it will also, if you ask it a yes or no question, it will give you a consensus meter and it'll say 67%, let's say, do retention programs work for retaining undergrads?

Well, it'll go through and it'll say, you know, 60% of the studies that we found said yes, 30% said no, and 10% had mixed results or something like that.

And it's pretty interesting.

Again, I don't know that I would 100% test it, but for a lot of things, that might be good enough.

So, Craig, when you run a query like that, does it give you some links to places where they would be in favor and not in favor.

So you can go and read those studies and make up your own opinion.

It does.

So it gives you a whole list of the studies and then it'll show you kind of red, green or yellow, red if it was no, green if it was yes, and then yellow or orange or something if it was unclear.

And then you click and you can read those studies for yourself.

And you can also ask it non yes no questions.

Then you don't get the consensus, but you get some results that are kind of similar to perplexity.

Perplexity and consensus both have free tiers.

So again, I would check it out and see which one kind of seems useful to you, but I like, I use perplexity a lot more than I use consensus.

But for certain types of things, especially those yes no consensus is pretty interesting.

So anything else that you use?

I've got one more kind of specialized.

Tool I've played around with a lot, but I wouldn't say they are things that I would use.

So why don't you tell us about what your specialized one is, Craig?

That'd be a good note to end on today.

So I'm going to, if you're hitting my keyboard, I'm going to make sure I got the right, I have the right URL.

So it's called Lex.

L e x.

Lex page.

It is a low distraction writing environment that has the best pre built prompts I've seen in a writing tool.

So for example, you write something up and you ask it for a critique.

It's got a little button called ask Lex and you can get feedback on your draft.

It'll flag confusing parts, brainstorm intro ideas, give feedback on an article idea, identify weak arguments, or overcome writer's block.

And then there's more.

And it's also got a chat window where you can just ask it a question like you would any other tool.

And it's almost like having an editor right at your fingertips, because it's, I don't know if you've noticed this.

The generative AI tools seem to be people pleasers, so they will be, oh, you know, this is really fantastic.

There's this one small tweak you should make.

And I, and Lex will say, yeah, this is a good start, but this part makes no sense.

And your transition here sucks.

And you've got this big hole in your logic.

It's not quite that blunt, but it gives you really good feedback.

Really, really good feedback.

That's quite insightful.

It's really like if you had an editor, you could turn your paper over to it also has a number of pre built checks for grammar, brevity cliches, passive voice, some things like that.

Its free version is so good that the only reason I went to the paid version is I want to support the project.

The free version is just fabulous.

I have not tried it for academic writing yet, but I plan to.

So this has been more for the newsletter.

Matter of fact, if you go to my AI goes to college substack next week, there'll be a review of Lex written using Lex, which is that what the kids call meta, or used to call meta?

Something like that?

I don't know, but it's really pretty good.

And like I said, it's this combination of this really well implemented AI with a low distraction writing environment.

And I cannot tell you how nice that is to not have a bunch of different commands and that kind of thing.

You're just kind of writing.

So do you do it all at Lex page, or does this plug into your word processing software and allow you to use it in say, word or pages?

No, it's all lex page as far as I know.

There could be something I haven't found yet.

You can upload documents, I haven't played with that.

So there's some other things.

And if you pay, they will actually schedule a 30 minutes onboarding with you so you can get on with the developers and they'll, I haven't done that yet, but, because, not sure I need to, but you can create your own reusable prompts.

Like, I have a terrible time with commas.

For some reason.

I way overused commas.

You know, I might build a prompt that says, just flag any place.

I use commas where I shouldn't have.

It's a bad spiral, I overuse them.

So I tend to try to cut back the use, and I always cut back the use of the ones I should have included, and it's not good.

Well, and what happens there, Craig?

And this is where I think editing and writing are two different things, is if as you're writing, you're trying to remember all your comma rules, you go from writing mode to editing mode.

And oftentimes I find that's what causes writer's block, is I'm too focused on the editing part of it that I don't write well.

No, you're exactly right.

If you write first and then revise, that's the way to do it.

But you know, it's free, why not try it out and if you like it, like I said, I decided to pay.

I think if you pay for a year, it's $12 a month.

It's 18 if you pay month by month.

But, you know, I thought it was worth supporting the project.

So check it out.

You can, you can do other things, like embed images and some other neat stuff in it, too, but it's pretty slow.

Cool.

I'll have to check that out.

All right.

Any final words, final thoughts?

I would encourage everyone to continue being curious, continue playing with these things and figuring out what you can do, because that's where you're going to begin to change the norm of how you approach searches, how you approach writing.

And I think it can only make you more efficient and better.

That's right.

Remember, you don't have to do 100% of your work.

If you can do 20% of your work, well, that's pretty good.

That's a date.

That's a day a week.

That's a day a week.

All right, that's it for this time.

Be sure to check out the newsletter and back episodes@aigostocolllege.com.

there's also a contact button.

We'd love to hear what your favorite tools are.

And that's it.

I will see you next time.

And Rob will see you next time.

Thanks, Rob.

Thank you, Greg.