Five months later, I still don’t want to use Google Bard

As we quickly strategy ChatGPT’s one-year anniversary, it’s protected to say that the world won’t ever be the identical once more. Over the previous 12 months, I’ve personally used the chatbot for planning holidays, fixing my good house issues, and the whole lot in between. And, for maybe the primary time in my complete life utilizing the web, a search engine has not been my solely supply of data or opinion. As a substitute, ChatGPT and Bing Chat have gained me and plenty of others over.

Due to a leaked inner memo, we now know that Google’s management realized the looming risk pretty early on and determined to reply in type. Shortly after, Google was able to retaliate with its first AI chatbot named Bard.

Google Bard was shrouded in thriller between its announcement and launch in March 2023, besides, everybody anticipated it to match ChatGPT. On high of that, its headlining function was that it may search the web to fetch correct responses — one thing you couldn’t do with ChatGPT on the time. All in all, it appeared as if Mountain View was just a few weeks away from dominating the net info area as soon as once more.

Quick ahead to right this moment and we’ve crossed the five-month threshold since Google’s chatbot turned publicly out there. Regardless of that, Bard hasn’t discovered the identical sort of success that ChatGPT achieved nearly in a single day. However at the same time as Google’s chatbot continues to fade from public discourse, the corporate hasn’t stopped engaged on it simply but. So after noticing the final set of updates, I made a decision to present the troubled chatbot a good probability.

Sadly, it solely took just a few exams to search out out why I finished utilizing Bard within the first place. I’ll minimize to the chase; right here’s a dialog the place I merely requested ChatGPT to record “the Galaxy S23 Extremely’s strongest benefits within the smartphone market”.

Do you notice something amiss within the above screenshot? In accordance with Bard, the Galaxy S23 Extremely begins at $1,179 and features a Snapdragon 8 Gen 1 chip and a 108 MP main digicam. All of these specs sound acquainted at a look however the keen-eyed amongst you might have already realized that none of these are true. The cellphone really begins at $1,199, consists of the newer Snapdragon 8 Gen 2 chip, and received a model new 200 MP sensor this era.

I initially chalked this as much as a one-off error, so I fed Bard the identical immediate as soon as once more in a brand new chat. This time, Bard responded with two right knowledge factors however continued to get the worth fallacious. I repeated the take a look at just a few extra occasions and located that a number of the drafts have been extra correct than others. However irrespective of what number of occasions I despatched in the identical immediate, Bard’s first response wouldn’t obtain 100% accuracy. I needed to both ask a follow-up query or discover a hidden draft with the right info.

It’s straightforward to see how this may be problematic. Think about this dialog from the angle of somebody who doesn’t know a lot about smartphones. When you depend on Bard’s responses to match units, you is perhaps misled into believing the Galaxy S23 Extremely has worse processing {hardware} than many different 2023 smartphones.

Keep in mind, we’re speaking about one of many highest-profile Android smartphones available on the market. A number of months have handed because the Galaxy S23 Extremely hit retailer cabinets, which means Google has sufficient correct info on the primary web page of its personal search engine. It’s attainable Bard would carry out even worse when introduced with a much less widespread gadget. Living proof: After I requested in regards to the Pixel Fold, Google’s chatbot insisted that the foldable’s outer show measured 6.7 inches as a substitute of 5.8 inches.

As for the way Bing Chat compares when introduced with the identical questions, it responded with excellent accuracy. What’s extra, Bing’s response completely laid out the processing and digicam specs that Bard received fallacious the primary time round.

Bard’s restricted information doesn’t simply finish at smartphones both. I’ve discovered that it’ll make up stuff about itself too, like within the beneath screenshot once I requested which language mannequin it’s based mostly on. It wasn’t till I identified the error from my very own reminiscence that Bard determined to confess fault. And even then, it claimed that the replace landed in 2022 earlier than the chatbot even existed.

Later in that very same dialog, Bard hallucinated but once more and started referencing a non-existent Google weblog publish. Asking for a hyperlink yielded no significant response. The screenshots above have been captured just a few weeks after Google’s announcement that Bard would use the corporate’s extra superior PaLM 2 language mannequin.

Primarily based on these outcomes alone, I’ve misplaced all inclination to belief Google Bard. I’d even go so far as to say that the above chats have solely bolstered my respect for ChatGPT and Bing Chat for delivering a secure and consistently enhancing expertise. Sadly, although, this has been a constant development ever since Bard’s launch. Assume again to when Google solely launched the chatbot in two international locations, with assist for only one language, and likewise prevented it from answering any coding-related questions by any means.

ChatGPT, in the meantime, surpassed Bard in all of these areas from the very first day of its unceremonious launch. Google didn’t even supply a chat historical past at launch, though that has since been added.

So why does this hole between Bard and its rivals exist? One issue may very well be that Google depends on its in-house PaLM 2 language mannequin, which can lack information in some areas in comparison with the fashions that energy ChatGPT and Bing Chat. However it’s additionally attainable that Google needed to minimize corners to fulfill investor expectations and rush its AI chatbot to market. We all know that Microsoft examined Bing Chat with a closed group of customers for a number of months, if not years, and has massively benefited from its investments in ChatGPT creator OpenAI.

Whenever you be part of all of those puzzle items collectively, it turns into clear why Google desires you to think about Bard as a artistic companion quite than a search device. The corporate expects errors and needs to get forward of them. Nevertheless, I don’t suppose that’s not an inexpensive expectation in any respect. Individuals will imagine Bard if it sounds assured even when incorrect, which it completely does proper now. There’s nothing Google can do to vary the general public’s religion in giant language fashions, wanting admitting defeat or enhancing its fashions in a single day.

In my view, I’d quite not use an AI chatbot than use Bard in its present state. I’ve come to belief ChatGPT a bit extra because the newest GPT-4 mannequin will usually a minimum of admit to not realizing one thing quite than pretending in any other case. And if I’m searching for probably the most correct info, Bing Chat provides me loads of supply hyperlinks to fact-check its responses. There’s merely no void that Bard can fill in my life and I don’t see that altering anytime quickly.