A Week with Bard — How it Compares to ChatGPT
Bard
Google just rolled out Bard for a preview and I had a chance to spend a week with it testing it out on various scenarios. I tested Bard for the same tasks I have previously tested ChatGPT for to really tell the difference between the two. Note that at this point ChatGPT has evolved into full release with free and plus subscription models. Bard, however is still in its infancy of adoption. My testing is not need based rather a like for like evaluation to help myself and others to decide which they eventually want to pay for. So I will try rely on my beta testing of ChatGPT
Now let’s see, how they compare..
Writing Tasks
Let’s take what I wrote above for introduction of Bard, and see what each of the two will give us
Bard
Bard gives you multiple drafts at once which is cool, but here Bard has tried to infer stuff I didn’t write in all three drafts
Google recently released Bard for preview, and I had the opportunity to spend a week testing it out in various scenarios. I tested Bard on the same tasks I had previously tested ChatGPT on, to really see the difference between the two. Note that at this point, ChatGPT has evolved into a full release with free and plus subscription models. Bard, however, is still in its infancy of adoption. My testing was not need-based, but rather a like-for-like evaluation to help myself and others decide which they eventually want to pay for. So I will try to rely on my beta testing of ChatGPT.
Overall, I found Bard to be a promising new language model. It was able to generate creative text formats, translate languages, write different kinds of creative content, and answer my questions in an informative way. However, it was not as accurate or comprehensive as ChatGPT. For example, when I asked Bard to summarize a book, it provided a brief overview that was missing some key details. ChatGPT, on the other hand, was able to provide a more detailed and accurate summary.
I also found that Bard was not as good at handling follow-up questions as ChatGPT. When I asked Bard a question about its summary of the book, it was unable to provide any additional information. ChatGPT, on the other hand, was able to provide more details about the book and its author.
Overall, I think Bard is a promising new language model, but it is not yet as accurate or comprehensive as ChatGPT. I would recommend waiting until Bard is further developed before using it for important tasks.
ChatGPT
ChatGPT however using the 3.5 legacy model produced below text. This was the case in the early stages of ChatGPT too, What I asked it to rewrite, it was good at doing just that.
Google recently launched Bard for a preview period, and I was fortunate enough to spend a week evaluating it across a variety of tasks. I compared Bard to ChatGPT, which has undergone substantial evolution and now boasts both a free version and a premium subscription model. My goal was to perform a side-by-side comparison, rather than an evaluation based on need, in order to assist others in making an informed decision about which model they may eventually choose to pay for. Drawing upon my previous experience testing ChatGPT, I sought to accurately compare the two models.
Coding
This is something every developer and would be developer, data scientist and analyst would care about. so here it goes
Task 1: Simple Bash command to replace a word in a file
Bard
Once again provides 3 drafts here, 3rd draft is shown in figure below.
Other two suggestions are below
ChatGPT
Plain and simple, gives one option, which you can certainly use to accomplish the problem. If you know what you want, you can further prompt it to generate other variations, Here in this simple task I like Bard’s performance providing multiple variants to test out.
Task 2: Python Coding Problem
Bard
Once again, Bard shines here with multiple examples of the code with 3 drafts.
Another nifty feature in Bard is inplace edit of your prompt as shown below, you don’t have copy paste it again with modifications like with ChatGPT, you can edit in place and regenerate. Depending on your needs can be very convenient.
ChatGPT
Provides a better explanation than Bard but again a single variant here.
Travel Suggestions
I never tested this in early version of ChatGPT, This was something I started with Bard, So I asked Bard for best hikes near Denver, As I live in Denver I can semi validate this.
Bard
Bard really shines here as one would expect with Google Maps being most used along all the location based comments people provide.
ChatGPT
Here ChatGPT gets pretty close with some good overlap but I like the list Bard provided definitely.
I also asked them both for suggestion on best vinyards in Bordeaux
Bard
Great suggestions, multiple Drafts with different explanations.
ChatGPT
Pretty similar to some of the drafts from Bard albiet a different list, Intutively I may trust Google here as it has the location data from its maps service.
Conclusion
In summary, I have seen Bard improve rapidly in the one week I have been using it. ChatGPT may have a headstart but given ubiquity of Google services such as search, youtube and their reach via Android, payment services and much more, Google in theory can be a great competitor to ChatGPT. It is yet to be seen how well this turns out for them, For Now, I’m happy to compare between the two as and when I need AI to augment my work.
On the other hand, with GPT-4 ChatGPT Plus subscribers are expecting to go beyond text into multimodal, What will that do to nascent Bard is also yet to be seen.
Finally, I love how much AI has pervaded everyone’s lives in a short period with the trendsetter award going to ChatGPT.
