Comparing apples with apples šŸ“¹

Let's look at things Bard could be good at - like website FAQs and YouTube titles.

On Monday, I started to put Googleā€™s new ā€˜Geminiā€™ Model through its paces, and now on reflection, maybe I was a little unfair.

As with every new employee, you should probably give them a job on the first day they have more chance of aceing (instead of failing!)

So today, I decided to put Bard, Claude and ChatGPT through their paces on a couple of tasks that Bard should be good at:

  1. Crafting YouTube Headlines (Bard theoretically should be good at this because Google owns YouTube, and the training data should be good)

  2. Creating a website FAQ (Bard should be good at this, too, because after all the years of crawling websites, one should be able to safely assume Google understands what a good FAQ looks like) 

So, here are my prompts

These prompts are pretty much a copy-paste of some posts on this blog from earlier this year, including Jab, Jab, Jab (in honour of one of Gary Veeā€™s first books), and on swiftly creating a website FAQ.

Here they are:

Provide examples of compelling headlines in the style of Gary Vaynerchuk for a Youtube video that offers tips on navigating the current [Broadbeach]real estate market. The headlines should encourage viewers to click and watch.

What are some hashtags?

Compile a list of frequently asked questions and their corresponding answers from customers of a real estate business in Broadbeach. This will be used as the FAQ section on the business's website.

Could you please do that again, but could you maybe adjust for a real estate business in Australia and SEO optimise for the suburb Runaway Bay?

The results:

  1. As expected, Bard did really well on the first prompt, and you can check the results here. What I liked about the result was how Bard broke down different headline suggestions into categories, for example, Clickbait, Intriguing, Action-Oriented, Benefit-Driven etc. The hashtags look pretty spot on as well and include some good tips for using them.

  2. With the FAQs, I thought again that Bard performed really well, and you can check the results here.

Whatā€™s nice about this new version of Bard (that I also didnā€™t pick up on the other day) is that there are several ā€œone-clickā€ buttons at the bottom of each response which allow you to quickly change the tone and length of the response, as well as a quick search-based fact check.

I made this quick video here to show you how that works:

And yes, the results for these queries on Bard were better (IMO) than Claude and ChatGPT at this point in time.

Claude was closest to Bardā€™s responses, with ChatGPT falling behind a bit.

When it comes to producing video headlines, ChatGPT doesnā€™t really look as though itā€™s moved far from my original post (backwards, if anything!)

Interesting times for sure!

Happy Hunting šŸš€

Reply

or to participate.