Hello Bard
So I got access to Google Bard.
This is a hefty lift for a large consumer and business-facing company and comes with a lot of attention from the media. Large Language Models are in the zeitgeist.
My first thought was to try out similar prompts I had used on ChatGPT. However, it’s not really possible to go beyond that with a side by side comparison because my responses are based on the AI’s responses.
After only about an hour of testing, there are clear differences and at least for the interim, either because I’m biased from my own sunk cost of learning or just more familiar with it, I’m preferring OpenAI’s. Bard seems like GPT3 or 3.5 in some of the errors of its responses and how it needs to be coaxed into trying to give an answer.
There some UI differences that I like. First, it gives the entirety of the response at once rather than typing it out like OpenAI. Second, they have made the response easily copyable, which helps. Third, it’s better at presenting tables and diagrams, especially ASCII art.
There’s still a lot to try and test but this is a good starting point and what we’ll likely see are two things: 1) the proliferation of many similar LLM-based generative AI tools and 2) rapid improvement.
Excited to see more.
