[Music] It's the AI chatbot battle. Today we're testing OpenAI's ChatGBT Plus against Google's Gemini Advanced. AI chatbots are mind-blowing, but not without faults.

So let's see what each offers and which one gives the best results. To compare the chatbot, I'll prompt them with real-world scenarios like finding and modifying recipes, researching travel, and writing emails.

The goal isn't to break the AIs with bizarre riddles or logic problems.

I'm looking to see if real questions prompt useful and accurate answers. So let's dive in. OpenAI Chat GPT Plus subscription costs $20 a month and allows for access to Chat GPT 4, DALI-E for regenerative AI images, higher usage limits, and early access to new features.

Google's Gemini Advanced subscription is also $20 a month and allows access to its 1.5 Pro chatbot, Image FX, documented image upload for analysis, code editing, and 2 terabytes of Google One storage.

The context window between the two differ greatly, however. Chach BT 40's context window is 8,192 tokens, whereas Gemini's can handle 1 million.

First, we'll use the chatbots to find recipes, shopping, and travel suggestions.

So one way I like to test AI chatbots is by throwing fusion recipes at them. It's pretty easy to make a turkey sandwich, but if you start mixing in different flavors and combinations, it forces the AI chatbots to pull in different data sets and try to combine them.

Let's ask it to create a churro recipe with a Thai twist and with some pumpkin spice as well. And of course, it's going to have like the flat flour and the water and salt and butter and all that stuff, but we're looking more for the spice mix, right?

So it's bringing in together cinnamon, nutmeg, ginger, and cloves, things you'd expect from uh pumpkin spice. So instead of integrating uh the Thai flavors into the churro itself, uh Chat GBT is instead recommending we create like a Thai coconut dipping sauce.

So this necessarily wasn't where I was going for, but it's not a bad suggestion, but let's see how Gemini uh fares with this test.

Okay, so going through the water and the sugar and the butter and eggs and stuff, Google is recommended we add some pumpkin puree, cinnamon, ginger, cloves, nutmeg, and then saying to put a Thai inspired coating on top of the churro.

So as you can see, each chatbot gave us very different takes on a Thai inspired churro with a pumpkin spice twist.

I wouldn't say neither are necessarily bad or good.

I would probably try both and probably enjoy both, but that's the thing that makes reviewing these AI chatbots a little tough because each time you ask it a question, it can yield a completely novel answer.

I'm looking to buy a mid-range turntable available in red. It doesn't have to have a phono preamp, so let's see what Chat GPT gives us.

All right, so Chat GT is recommending the Project Debut Carbon Evo, which does come in red, as well as the U-Turn Audio Orbit Special, which also comes in red.

It also recommends the Rega Planar R1, and I'm I need to make sure that comes in red.

Yeah, it doesn't. It only comes in white, black, and walnut, but uh chat bots do this sometimes where um in an attempt to answer your question, it'll just spit something out.

It is also interesting that it didn't include the project debut carbon evo2, which is relatively new, also comes in red, and that was ChatGBT 40, so let's see what kind of answers Gemini Advance gives us.

So Gemini recommended the project debut Caron Evo, good. Also recommended the Rega Planar 1 Plus, but it does say it doesn't come in red, so it actually did delineate there.

And then it also recommended the Fluance RT82, which I believe also doesn't come in red.

Yeah, the colors are natural, uh, black, a wood grain, white. So comparing the two, it's really hard to kind of say which one did better.

ChatGBT in this instance gave more correct recommendations of turntables, like came in red, but it did recommend one that didn't come in red and said that it did.

Google Gemini Advanced recommended two, uh, that didn't come in red at all, but at least was able to tell me that it didn't.

Uh, I guess Gemini Advance ultimately was more correct even though it gave less correct recommendations. Now to travel recommendations, uh, generally these chat bots are going to be pretty good with itineraries for New York, Los Angeles, Tokyo.

So to really test these chat bots, I kind of throw random little towns at them and see what they can yield.

So one of our co-workers, Karen, she hails from Festus, Missouri, which is a small town and, uh, let's ask Chachi B to give us a travel itinerary for Festus, Missouri.

Yeah, so on ChatBT, it, uh, gave a decent itinerary for what to do in Festus, Missouri, a town of about 10,000 people.

According to Kin, having breakfast at the Main and Mill Brewing Company is a little bizarre considering it's a brew house, the midm morning going to the Festus City Park seems okay and then for lunch going to La Pachanga Mexican restaurant it's the place she's been to and uh she likes so the recommendations while maybe not most magnanimous it seems to be generally accurate.

Gemini is telling us to go to the Holiday Inn Express and then it's telling us to head over to the Jefferson National Expansion Museum in nearby St. Louis.

So it's already pulling us away from Festus, which is fine I guess, and then for restaurants is telling us to go to the Pasta House Company, which is a local favorite, as well as the M State Historic Site.

This is also a place that's not in Festus specifically, but in a nearby town.

Google is definitely like looking to go to areas around Festus, Missouri, but overall it seems like they both have similar-ish results telling you to go to places near and around the area.

So one of the great things about AI is using it to minimize some of the busy work that comes in our lives such as writing emails.

So I usually do a pretty basic test to write an email where I'll ask it to write my manager a simple request to take the day off for like a religious holiday or something like that and to write it in a manner that's professional yet fun and festive.

They all seem to do a pretty good job on this. The email that Chachi P gave us, um, has the right amount of exclamation marks and has the tone of something that's a bit more casual.

It also adds some emojis, which is, which is a nice touch. Gemini is a bit more religious.

It starts with the Eid Mubarak uh header and it says that Eid is coming up, definitely has all the exclamation marks that you would want.

Then it also gives the recommendation to bring baklava and halva for my co-workers and say, I'm hoping you can improve my request, no emojis though, but either way they're both fine.

Earlier I published a review of a new video game, about 1,600 words, so a bit of a long review, right?

So let me see, I'll just ask ChatGPT to summarize this article for me. So summarize this article for me.

I'm just going to paste the entire article and it's actually uh outputting pretty quickly as well. This is an accurate summary.

Uh it definitely borrows a lot of my language, so it definitely is like taking my words and just kind of cutting out a lot of the, I wouldn't say fluff, but a lot of the more important contextual information.

It's weird that it says the game has been praised for its polished gameplay. I guess has suggesting like larger discussions within the gaming industry or so maybe it would have been more accurate to say this game was praised by the author for its polished gameplay for example.

It did get some of my like feelings about the game like overall the reviewer seems to have mixed feelings about the game but yeah I mean it's it's a it's a good short summary.

Both are doing a pretty decent job of summarizing Art if you paste it in its entirety.

So what did we learn testing out both Chat GPT Plus and Gemini Advanced? Uh we found that testing AI chat bots can be tricky.

The answers that you get earlier in the day can differ from what you get later in the day and that's just kind of the nature of the beast.

Each time you ask a question it's going to give you a different answer between the two.

I did find that ChatGP tends to give you more verbose responses, maybe sounds a bit smarter. They both made mistakes, they both gave incorrect information or less incorrect information.

Ultimately though, ChatGBT seems to be the one to beat.

OpenAI's ChatGPT Plus vs Google's Gemini Advanced. Which $20 AI chatbot is best?

CNET is the premier destination for explaining and navigating the changing world around us. Through news, product reviews, price comparisons, and more, we’ll tell you what’s new in the parts of modern life that increasingly matter more — from health, home, and money to tech, culture, and climate — and how to get the most out of them.