AI Apps - Gemini Pro First Impressions

Trying Gemini Pro out on the Pixel 8 Pro

Google Gemini AI app on Pixel 8 Pro

Gemini Pro is the latest (allegedly greatest) AI app I’ve added to my Pixel 8 Pro. I’ve been using it for a week now and have some first impressions to share.Some of them are impressions of the Gemini app, others are comparisons with the other AI apps (or chatbots) I use regularly. Here are the six I am currently using - the widget underneath the weather widget is Perplexity, below it left to right are Microsoft Copliot, Gemini, ChatGPT, Pi, and Poe.

Perplexity allows me to select from four different large language models when using it, and Poe allows me to choose from more than 20.

AI apps on Pixel 8 Pro

Gemini Pro - The Good

The good things I’ve noted about Gemini Pro in Week 1 of using it are:

• It pulled in my chat history from when I was using it (when it still had the Bard name) as a web app.

• When I’m using Perplexity and choose to use the Gemini Pro model, I can view that same chat history.

• Using Gemini Pro as a replacement for Google Assistant is mostly a positive experience:

  • It’s nice that it can be launched using various gestures (I use a double-tap on the back of the Pixel 8 Pro).

  • The combo of what Assistant was when it was at its best plus a chatbot seems promising - although half-baked at the moment

  • It recognizes when you’re finished speaking and starts responding without the need to tap anywhere to kick that off.

• It has a Privacy Hub.

Gemini apps Privacy Hub

The Could Be Better

Gemini Pro makes one simple - and often used - task a little slower than Google Assistant. With Google Assistant, you can invoke it, say “Take a screenshot” and it will immediately do that. With Gemini Pro, you need to first tap on the microphone icon, then tell it by voice or typing to take a screenshot. This is a small thing, but if you’re like me and you do this several times a day, it might become annoying.

Another bit of bad news on the screenshot front with Gemini is that it includes its own prompt/command bar with the words “Take a screenshot” in the screenshot. Assistant was smart enough not to do that. Again, this is a small thing and hopefully a bug that will be squashed in an upcoming update to the app.

A much more goofy could be better item is that Gemini Pro does not add links to its sources to its responses, as Perplexity, Copilot and others do. You need an extra tap on the Google Search button to do that. Another small thing, but considering that Google is the 800 pound gorilla of web search, it feels odd and inelegant.

Gemini Pro vs The Competition

Gemini is fast. It starts, and finishes, responding quickly. It often delivers what I consider an appropriate length of response without specifically telling it to do so within my prompts. It’s also strong on accuracy of responses, offering very few inaccurate or off-topic answers.

Compared to the AI apps from OpenAI (ChatGPT), Microsoft (Copilot), and the other four I mentioned above, Gemini is already a worthy competitor. It feels as fast as any of them except Perplexity, and as just as accurate.

Having said that, other than Perplexity’s speed and unique features, there’s not a whole lot of separation just now between these apps. They have very similar feature sets, their UIs have a lot of common elements, and none of them are lapping the field in terms of performance and quality of responses. I expect that over coming months there may be more separation between these, as more money is pumped into the models that power them, the apps are updated, and maybe some new rivals emerge.

I may have some further thoughts to share on Gemini Pro’s big sibling soon. I just got access to Gemini Advanced this afternoon.