ChatGPT Inside Google Sheets
Toughts & Comments
ChatGPT Inside Google Sheets - YouTube https://www.youtube.com/watch?v=eV59wXCTy2M Transcript: (00:00) Hi everyone. Michael here. Today, we're going to take a look at how you can use AI inside of Google Sheets right now. You can do some pretty amazing things like write complex formulas and then have it explain it back to you in simple English. You can also do things like summarize text and analyze sentiment. Let's go take a look. (00:21) Today, we're going to be using a third-party tool called Numerous.ai in conjunction with Google Sheets. It's free to try out and use, but if you use it a lot, it will cost some money. You can download it up in the top right-hand corner or down below in the description. Here we are on the Numerous.ai homepage. There are a lot of tools coming out right now that offer this functionality, but I like Numerous. (00:43) ai because it handles so many different scenarios very well, and for some reason, Google doesn't offer this functionality yet itself. You can scroll down on Numerous.ai's homepage and check out additional scenarios that you can use their tool with, but for now, I'd like us to click install in the top right-hand corner. (01:00) Now that we've installed the Numerous.ai add-on, we can use the functionality within Google Sheets. It'll query ChatGPT directly. If you want help with the functions, which I usually do when I start using a new add-on, you can go up to the top toolbar here and select extensions, then scroll down to Numerous.ai and select show sidebar. The Numerous. (01:24) ai sidebar is really helpful because it gives us prompts to use when we're generating functions. I'm actually going to walk through some of these functions right now to help me get my work done here for the Kevin Cookie Company. First thing I'm going to do is close the sidebar, even though I just said we'd open it. (01:40) I kind of know what I want to do right now. I want to create descriptions for cookies off here on the left. My boss has asked me to come up with some body text that includes a description for each of these cookies, along with its corresponding price, and I don't want to do this manually. (01:56) And that's probably why you're here in this video, too. You don't want to do all of this manually. So I can actually write a formula to do this. The formula here will be equals write, which is one of the prompts that we saw for a function over here. And I'm telling it to write a description about this cookie and the cookie being in A8, in this case chocolate chip. (02:17) I want you to write a description in an enthusiastic tone in one sentence and include the price listed here in B8, which will be $5.50. And you can see that that is exactly what ChatGPT has done for us. And it is pretty, pretty good. And there you can see the function inside of it working as intended. But you could go out and then control, copy, paste as values if you want. (02:44) And that way you won't have the formula anymore. But I'm not done with the formula yet. I actually want to expand this down to the bottom and see if it will do it for all of our cookies. It takes just a second to load, but I promise you it's going to be worth it. And there you have it. (03:01) Oatmeal raisin, snickerdoodle, all covered. And let's double check to see if the prices are included. Says here row 15, $2.25. Let's double check that. Yep, $2.25. Everything seems to be matching up nicely. While our descriptions are pretty helpful, there are a few errors I noticed when reviewing them at first. You can see up here that it says 0. (03:27) 5 and here it says two and then the word dollars. And even down here at the three, the dollar sign is behind the three, and I want that in front of the three. And this is just a great example of how ChatGPT is really helpful, but still imperfect. You're going to want to check your work all the time when you use these kind of AI assistant features. (03:51) You might be wondering why you can't just go directly to ChatGPT and come up with descriptions there. You could. These would have to be entered in one row at a time and you'd have to specify the price each time. This is helpful because you can do it in bulk and it is the power of Google Sheets and bulk editing and data management altogether paired with ChatGPT, and it just saves you a lot of time and headache. (04:12) And now that we've wrapped up filling in all the descriptions and their prices, I still do have some work left to do for the cookie company. So, we're going to move on to our next example. My next assignment for my job is to find out the names of our customers who sent us some feedback. I want to parse their names out from their customer email addresses so that we can give them a more proper greeting when we write back to them. (04:39) To do this, I'm going to want to generate a formula and to find out how to do that, I will go up to extensions here in the ribbon toolbar, scroll the numerous.ai, select show sidebar. Once it has reappeared, I will go to generate formula and I will describe in plain English the formula I would like this tool to generate. So, I will give that a try. I say extract everything to the left of the at symbol in column A. Let's see what it does. (05:06) It has produced a formula for us. It says equals left A:A. All right. Well, let me give this a try. Copy it and I will paste it over here. Or you could alternatively just hit insert into cell. Maybe I'll try that. And look at that. It did it. Over here we have Kate at KCC.net and here it populates her first name, Kate. (05:30) It's wonderful that the formula worked, but I don't fully understand what the equals left formula means. So, I'm going to copy it and scroll down to explain formula in the formula here and click explain formula. It says this formula returns the leftmost characters in column A up to the first @ character. And that's actually what we wanted it to do. And I find this to be a very helpful and simple basic explanation. (05:58) Next to expand the names down, I will click right here, double click actually, and we can see that the formula has worked. As I mentioned earlier, you can select all of these names and do a control copy and then paste it as values so you can get rid of the formula if you'd like. My next task is figuring out whether the feedback we received from our cookie company customers is positive or negative. (06:21) First, I will close the sidebar. I'm going to train the model we're going to use next. So, I'm going to have to do a couple of manual reviews first. The first piece of feedback says I was disappointed with the size of the cookies I received. (06:37) Well, I don't know what they were expecting, but I suppose I should jot this down as negative feedback. Just below it says I recently tried Kevin's chocolate chip cookies and they were amazing. OK, what we like to hear, positive feedback. Next is where the magic is going to happen. I will go back up to extensions, go here, show sidebar, and I'm going to check to see what I want to use and I want to use infer. (07:00) This is what we're going to use next. Yep. This is where you can look at sentiments of things, positive or negative. Yep. Double checking. This is what we want. I'm going to write equals infer. And it says to target some example inputs, the example inputs in this scenario are the blocks of feedback. The example outputs will be the negative and positive sentiments here. (07:26) So, I'll select those two and then it will say the target input, which will be this unscanned block of feedback, and then I will put a parentheses around it and run it. It says loading and it has quickly figured out that I love that the Kevin Cookie Company uses these high-quality ingredients, and their cookies is indeed positive feedback. (07:51) And now I'm going to fill in the rest of the rows in column D10 by double clicking right down here. It's starting to approach the end of the day here at the cookie company, and my boss has asked me if there are any standout pieces of feedback that they should be aware of. (08:07) I've scrolled down and noticed that a user or a customer rather has left a large block of text for me to digest. I want to see if ChatGPT can summarize this for me into some bullet points. So, I'm going to use one of their summarize functions. You can see up here I've asked the AI to summarize this into three bullet points and this being all of the text in cell B15. (08:31) It has done a pretty good job. This is the cookies arrived stale. The customer service was apathetic and also a positive experience was not achieved. So, I do think ChatGPT did a really good job of summarizing this for me. (08:49) There's one last thing I want to touch on before we end here today, and that is that everything we've done here so far has been completely free. But if you want to use Numerous.ai quite frequently or want to use a lot of queries against ChatGPT, they do run on what's called a token system, and they start you off with 60 free tokens. Each token is used when you create a new function, and if you want to have more tokens, you can always choose to upgrade here. But I want to reiterate, you can do a lot of really interesting stuff for free. (09:20) Wow, that is pretty groundbreaking stuff. I don't have to remember every Excel formula or Sheets formula I want to use. There are so many assistants now that can help me do it. I really wish these things were around when I was just starting out. (09:35) And in the future, I don't know if we're going to need third party integrations, as I think Google and Microsoft are going to incorporate these into their own products. But for now, this is really useful and I suggest you try it out. For more content like this, please consider subscribing and we'll see you in the next video.