Bard is built for research, whereas ChatGPT is the better writer.
Investigar Google Gemini.
Let’s start with this: both Bard and ChatGPT will run into logic issues. For example, I asked them both the same simple logic question, and they both got it wrong.
GPT-4 (using ChatGPT Plus) got it right, but it still messes up plenty of other things. Bottom line: it’s pretty hard to compare the accuracy of the outputs—especially since a lot of it depends on how you phrase the prompt.
But lack of reasoning skills aside, Bard’s output is pretty helpful if you’re looking for succinct answers around any topic. That’s because Bard finds the most relevant information across Google and summarizes it for you. The benefit here is that you don’t have to click through different pages or compare information, as Bard will do that for you.
As part of ChatGPT, it functions as a chatbot, but it can also serve as a summarizer, a translator, and other roles on a more textual level.
This makes GPT (in my opinion) a much better writing tool than Bard. As a quick example, when I asked GPT-4 to write a tweet about Zapier, it did so pretty much in line with the rules for writing tweets (e.g., short character count, emojis, hashtags).
I found that ChatGPT is also better at brainstorming blog ideas, writing long-form articles or emails, and coming up with content marketing ideas. When I asked ChatGPT to provide me with an outline for a blog post about the crisis surrounding bees, the depth of detail in its output far exceeded Bard’s.
So, while GPT can understand and generate a wide range of text for multiple purposes (including content marketing), Bard feels like it was designed primarily to act as a research tool.
I asked Bard about recent discoveries made by the James Webb telescope, and it instantly parsed these results:
Speaking of hallucinations, Bard’s also prone to muddling its responses. When I asked about a book that was recently published by Kate Morton, this was its response:
If you’re using ChatGPT, the only way to get it to summarize articles for you is if you copy and paste the text from the article into the prompt box. AndChatGPT has a limit of 4,096 tokens, which is equal to roughly 3,000 words. For anything longer (like research papers), you’d need ChatGPT Plus.
Again, it generated a chunky response, which might not be ideal if you’re looking for a quick overview. Bard knows automatically to give you a quick summary in the form of bullet points, ideal for scanning.
However, you’ll see that ChatGPT was able to capture certain nuances Bard didn’t, like the fact that the toy market is shrinking due to birth rates dropping. This is pretty important if the beginning of a study dictates the reasons behind the rest of the topic, so it’s good to bear in mind that ChatGPT takes in the finer details.
And remember: you can specify how long you want your summary to be in your prompt, so even if your first attempts are a little wordy, that shouldn’t stop you from using it as an effective summarizer.
Surfacing images from the web for visual context
Bard’s ability to surface both search results and images from Google Search definitely sets it apart as a major player. Although it can’t create images using AI (yet), it can surface images that already exist according to the topic you’ve asked it about.
Whether you’re researching specific dog breeds, Vermeer paintings, or even bicycle repairs, Bard can churn out specific images from other web pages for visual context.
For example, with the Expedia plugin, I just tell ChatGPT about a trip I’m thinking of booking, and it’ll immediately surface the cheapest flights it can find via the travel site, along with the link, airport details, duration of flight, and most importantly, the pricing.
It’s so much easier than going through travel sites yourself, adjusting filters, and comparing sites side-by-side. It will also offer lodgings or other area-specific activities for you to explore.
You can ask Bard to do the same, but it will provide you with completely made-up prices. And in this case, the images it provides (logos of the airlines) are just distracting.
Google Bard: pros and cons
Bard has internet access powered by Google Search built into its tool for free—providing fast responses
Bard is prone to hallucinations, so everything has to be taken with a pinch of salt
Bard is better at surfacing relevant information (including images) from Google Search.
Sources are not always reliable and should be fact-checked
Bard has a more user-friendly interface, with nicely formatted (human-like) responses
Bard provides a fairly isolated experience, with no plugins or integrations
ChatGPT: pros and cons
ChatGPT is better at generating text (like creating long-form content)
ChatGPT Plus has access to a web browser (powered by Bing), but it’s a separate experience and can be slow at times
ChatGPT is a more collaborative experience, with the ability to share conversations with others
ChatGPT’s responses are often quite lengthy and the text chunky, making it hard to scan
ChatGPT has a whole suite of plugins (and a Zapier integration) that offer more use cases with different apps
ChatGPT is also not immune to hallucinations and poor reasoning, so must be fact-checked
The better AI tool depends on what you’re using it for—and whether you can come to terms with those pesky hallucinations.