Pitting AIs against each other to do my work, working harder
I decided to attempt to recreate my MoGo 2 Pro article using AI. To these ends I pitted the Bard, ChatGPT, and Bing against each other and asked them to do the same thing. Here’s what I got.
TL;DR – short attempts to get good data fail
ChatGPT
Of the three specs ChatGPT listed, only one was correct.
Attempting to get some clarification on it it added new incorrect information indicating that it had a 10 amp battery (it doesn’t) runs Android TV 9.0 (it’s Android 11)
Formatting nice, information useless.
Bard
Almost the same set of incorrect information as ChatGPT – it’s pulling some of the MoGo Pro specs and mixing them in.
Bing
While both the other AI chatbots got the lumens wrong, Bing at least linked to the location that listed the incorrect information. Oddly it lists both the incorrect lumens and the correct lumens.
Asking about specifics it generally returned the right information.
It alternately told me there is no battery, and that it needs a 65Watt charger (correct,) or that it has a built in battery capable of lasting 4 hours (incorrect).
Bing shut me down when I asked for specifics at one point, but allowed me to ask the same question in a new chat.
Nicest formatting.
I asked several follow up questions, noticed that Bard takes the fact that it’s wrong and admits it and then forgets the right answer. I figured maybe this is just because the MoGo 2 Pro is new and we’re missing a lot of data let’s look at something older. In my case I decided to look at the Casio SK-1 Sampling keyboard.
Bing gave me a short, but completely correct response, Bard gave a factually correct page long rundown of the thing, and ChatGPT provided a nice looking rundown with two glaring errors.
