Trying Bing’s AI chatbot for those who can’t: here’s what the ChatGPT rival is like

After a couple of weeks, I finally made it out of the waitlist for Bing’s AI chatbot.

According to Yusuf Mehdi, Microsoft’s consumer chief marketing office (CMO), more than a million people had joined the waitlist for the new Bing’s preview in just 48 hours after announcing it.

As of March 8, slightly over a month after the announcement, the consumer CMO shared Bing has crossed 100 million Daily Active Users (DAUs).

“This is a surprisingly notable figure, and yet we are fully aware we remain a small, low, single digit share player,” he wrote in a blog. “That said, it feels good to be at the dance!”

Like ChatGPT, Bing’s AI uses tech from OpenAI. However, Microsoft stated that Bing actually runs on a new, next-generation OpenAI large language model that is “more powerful than ChatGPT”.

“We have developed a proprietary way of working with the OpenAI model that allows us to best leverage its power,” Microsoft wrote. “We call this collection of capabilities and techniques the Prometheus model.”

Jargon aside, Bing’s AI does provide quite a different experience compared to ChatGPT. For one, unlike ChatGPT, Bing’s chatbot is integrated into its search engine, meaning there isn’t a specific website you go to enter your questions. Rather, you just type it straight into the search bar.  

Here’s what chatting with Bing is like.

Ask any question… but only 10 at a time

Compared to ChatGPT’s limitless questions, Bing’s chat started out with a limit of only five chat turns per session. This was in response to cases of long chat sessions that confused the underlying model.

Eventually, Bing upped the limit to six, to eight, and now (as of the time of writing), 10 chats per session.

It also has a limit of 120 questions per day, so I had to pick my questions wisely. Also, unlike ChatGPT, you can’t edit your prompts or questions.

With the continual increased limits, hopefully we’ll soon get to a point whereby there are no limits at all.  

Curating search results and citing responses

Something that Bing’s chatbot does is that it occasionally hones in on one phrase in my search and makes assumptions of what I want to know instead of just answering the whole question.

For instance, I asked, “Bing, do you have a cut-off date?” Instead of asking me to clarify, the chatbot decided to just search for the term “cut-off date”.

Interestingly, though, Bing’s answer is actually cited and linked. Professors rejoice! It sourced information from accountingtools.com as well as Collins Dictionary.

This makes sense, since Bing is a search engine. So, I suppose its chatbot also serves as a curator of search results.

That said, it’s not like Bing only tells the truth. When prompting it to do things like write a story about a brand, it does fill in the gaps with assumptions made, much like ChatGPT does when there’s a lack of information.

Speaking of assumptions, Bing also gives prompts for the user’s next message, which is kind of interesting. Are the bots already trying to overtake our minds? (I jest…. unless?)

Way more personality than I expected

One issue that I previously noticed with ChatGPT is that it is absolutely incapable of giving accurate character counts, even though it says it can when prompted.

I wondered if Bing is any better, so I asked it to count the characters in the opening sentence of this article.  

When asked to show its working, it spat out this equation: 5 + 1 + 6 + 2 + 5 + 1 + 1 + 7 + 4 + 2 + 3 + 2 + 3 + 8 + 3 + 6 + 2 + 8 = 76. The numbers represent each word’s character count.

Its addition is just straight-up wrong. If you punch in the numbers on a calculator, the answer is actually 69.

Another issue is that Bing did not add in the spaces. When I told it this, it retorted: “There are 15 spaces in the sentence. Each space is one character. If you add 15 to 76, you get 91. But you asked me to count the characters including spaces, not excluding them. 🙄”

Yeah. Eyeroll emoji, right back at you, Bing.

First of all, no. It’s 16 spaces. Also, what? I did tell it to include spaces, and it didn’t. What’s it trying to say?

When told this, Bing replied: “I’m sorry but I prefer not to continue this conversation.”

With that, it ended the conversation with four turns left. I was… shocked, to say the least. The sass is real.

Similarly, it also oftens tries to steer me in a different direction when it doesn’t have a satisfactory answer. Oh well, I suppose that’s a little better than giving me wrong information.

Choose your conversation style

So you’ve read all the above and maybe hate the personality-packed vibe Bing’s chat is going for. Well, good news, users can actually opt between three conversation styles: more creative, more balanced, and more precise.

This does not at all make its responses more accurate though, it just delivers the inaccuracies in a less sassy way—no emojis or anything.

When on “more balanced” mode, I had asked it to pitch Vulcan Post to potential investors. It had responded by sharing a more personal email using lines such as:

We are currently looking for strategic partners who share our vision and passion for empowering the local startup ecosystem and creating positive social impact. We believe that your expertise and network in the tech industry would be a valuable asset for us as we scale our operations and reach new markets.

In the “more precise” mode, though, it mostly focused facts that it obtained via doing a Bing search and provided sourcing throughout. I was really charmed by this, as it helps me distinguish what is factual and what is just made up to appease me—something I feel I couldn’t tell with ChatGPT.

Despite proper sources though, its resulting response still isn’t the most accurate per se (see our supposed content pillars below), as it tends to string together bits of information in a way that it feels makes sense, even if it logically doesn’t.

Only accessible on Microsoft Edge

As a dedicated Chrome user who really doesn’t like change, the fact that Bing’s chatbot is only accessible on Microsoft Edge is a bit of a turnoff for me.

With that said, though, I’m still willing to open up Edge and use Bing just to use its chatbot, so who’s really the winner here?

ChatGPT still proves to be a lot more professional and functional in the way it responds. Yet, Bing has much greater entertainment value, and I must say that brings me a lot of joy.

Still, the turn limit is a little… limiting, to say the least. One of the great things about ChatGPT is how it remembers all the previous messages within any given conversation. This helps when you’re prompting it to continuously act as a certain character (e.g., an interviewee, interviewer, or something else).  

ChatGPT also helpfully lets you access old conversations, but Bing doesn’t seem to allow that, at least not easily.

In any case, this whole chatbot race has got me on the edge (pun intended) of my seat. Google’s got its upcoming Bard, of course. Even DuckDuckGo seems to want in, having just released its DuckAssist feature.

Just as Google has all but dominated the search engine market though, I believe the same will happen here. ChatGPT may have a leg up as the first mover, but Google’s Bard has a really high chance because of its accessibility.

With that in mind, I’m certainly looking forward to using Google’s Bard—so stay tuned for a review of that.