My first impression of Google Bard
I have read a lot about the new Google Bard, which should be the big challenger for ChatGPT. But I didn’t have a chance of trying it out myself and I don’t want to write stuff I didn’t do or saw with my own eyes. So, here is my first impression of Google Bard.
Before you continue; Note that this article has been written on 29 March 2023. Things go crazy fast, especially when it comes to AI. Some things in this article may not be valid when you read it in a week or two. I will try to keep this article up to date.
And I Am On A Waiting List
I saw this cool video of some guy telling us that Google Bard is finally here and we should use it… How? He did forget to mention a few things. Firstly, Google Bard wasn’t launched in my region/country. But with some little trick, you can be put… On a waiting list. What? That is the second thing this person forgot to mention. But okay, I put myself on the waiting list.
I did the same thing for Microsoft AI in Edge, but that took 4 weeks! Google seems to be a bit quicker. I filled in the form with some random reason that wasn’t really convincing. But 3 days (!) later I go the e-mail I could check out Bard.
Google has been really good with the Google Accounts and that works really well with Google Bard too. No endlessly long forms, just your Google account, and… GO!
I Am In!
And it looks so empty. Almost as if I just walked into a hospital, so sterile. But okay, it’s experimental (Google already told me that like 10 times), so maybe the design is just not there yet.
Because ChatGPT was first, I look for similarities, which is something you shouldn’t do. This is Google Bard, not OpenAI ChatGPT.
The menu is pretty simple, will click through there soon. The top is just a standard Google toolbar. Cool.
Let’s search for something. This afternoon I was busy with writing the 3rd part of learning C# and it was about C# classes. Let’s see what Bard can tell me about that.
Hey Google! Tell me what C# classes are in C#
The first thing I notice is… No error connection, no long waiting. I was almost about to get up to get some coffee. I mean, ChatGPT is writing back as if it was a real human… Google Bart clearly is not.
Much Quicker!
I just lifted my butt off my chair and BOOM! Result. This is brilliant! No long waiting for some machine to type the result! Or better yet: No long waiting just to see it ‘lost connection’ somewhere around the middle of the result.
The information is pretty accurate. But I find the information a bit too complex. A bit of information and a lot of code. My articles are targeting people that have no clue how to develop in C#. So I asked Bard to tell me more and treat me as if I have no clue what C# is. The result… Well, check yourself:
Again, very quick. No waiting (just a few seconds) and very good. This is something I could really use. No code and a lot of theoretical information.
Learning Languages?
Okay, I want more! So I called my wife, who was busy on writing a new article for her work. She told me that ChatGPT doesn’t have all the information and sometimes doesn’t know something exists while we know it’s really there.
She asked ChatGPT the following question:
“is there a maze at Parque de São Roque”
And this was the response:
We didn’t ask for any real-time information, just if there is a maze at some park.
Let’s ask Bard:
Okay… So that is useful to know, but it’s just English. Let’s ask it differently. First ChatGPT which took me 5 minutes to ask and take the screenshot.
And now Google Bard:
So, who is right? Google Bard is: https://visitporto.travel/pt-PT/poi/5cd04b45f979e000018e5ac1#/
Now, it’s not about who is right, but who gives the correct information. Oh, I think Bard will give incorrect information too. Maybe ChatGPT will have an answer when Bard doesn’t have one. But it’s kinda amazing.
Some Questions
A few things that I find inconvenient in ChatGPT are that it’s not real-time and it can’t connect to the Internet. What’s Bard’s stand on this?
So, I asked Bard if it can connect to the internet. It can’t, but the information is gathered from the Internet.
Another question I asked was if it has access to real-time information: Nope, it doesn’t.
One of the major reasons I use AI is to quickly get a piece of code I can use in my software development. I am a senior software engineer and it’s easier to do it by AI I can distinguish bad code from good code.
I asked ChatGPT and Google Bard the same question:
Can you generate an HTML page with CSS that looks like the Azure DevOps dashboard?
This is the response of Bard (let’s do this one first since ChatGPT is still typing…):
And this is ChatGPT:
A lot of HTML and CSS at the bottom. It doesn’t look pretty, but it’s something.
Conclusion
I will most definitely keep an eye on Google Bard. It’s fast and accurate. But I think a lot of information is missing. As if ChatGPT is a teen rebelling (freezing, connection lost halfway, running away (not available), and more) and Google Bard just knows how to walk.
Although I will continue using ChatGPT for code, I will be returning to Google Bard once in a while, seeing it grow and say its first words.
Yes, I am a Google fan, and most apps they have work just perfectly for me. I have no doubt this AI will be a great tool too. But when this will be? I have no idea.
