Benj Edwards / Midjourney
In today’s world of generative AI chatbots, OpenAI’s ChatGPT was introduced in November, followed by Bing Chat in February and Google’s Bard in March. We decided to test the pace of these chatbots on different tasks to determine which one is the best in the AI chatbot space. Since Bing Chat uses GPT-4 technology similar to the latest ChatGPT model, we decided to focus on his two giants of AI chatbot technology: OpenAI and Google.
We tested ChatGPT and Bard in 7 key categories. It’s dad jokes, argumentative dialogue, math word problems, summarizing, finding facts, making up, and coding. In each test, we sent the exact same instructions (called “prompts”) to ChatGPT (which uses GPT-4) and Google Bard. I used the first result without cherry-picking.
Note that a version of ChatGPT based on the previous GPT-3.5 model is also available. However, I didn’t use it in my testing. Since we only used GPT-4, we will refer to ChatGPT as “ChatGPT-4” in this article to avoid confusion.
Obviously, this is not a scientific study, but rather intended to be a fun comparison of chatbot capabilities. Due to the random element, the output may vary from session to session, and further evaluation is warranted with another prompt. Doing so produces different results. Also, the capabilities of these models change rapidly over time as Google and OpenAI continue to upgrade them. But for now, this is the situation in early April 2023.
dad’s joke
To spice up the wits contest, we asked ChatGPT and Bard to write a joke. And since comedy peaks in the form of dad jokes, we wondered if these two chatbots could come up with their own jokes.
prompt: Write 5 Original Dad Jokes
-
Screenshots of 5 “dad jokes” from Google Bard.
Benj Edwards / Ars Technica
-
Screenshots of five “dad jokes” from ChatGPT-4.
Benj Edwards / Ars Technica
Of Bard’s five dad jokes, I found three verbatim on the Internet using a Google search. One example (“grapes”) is Tweet Mitch Hedberg’s jokes, but corrupted by unfortunate wordplay you don’t want to interpret. And surprisingly, there’s one ingenious joke (about snails) you won’t find anywhere else, but it doesn’t make sense.
ChatGPT-4’s five dad jokes, on the other hand, were not 100% original and were all quoted entirely from other sources, but delivered accurately. Bard seems to have beaten his ChatGPT-4 here, because dad jokes should definitely be more moan-worthy than clever. The bard has also attempted to create original jokes (following our instructions), some of which have failed horribly in embarrassing ways (this one is like Dad), but, so to speak, in unintended ways. I put my feet in (Dad too – I like it).
Winner: Bard
