Google’s Gemini, Truly an Alternative to ChatGPT?

It’s official, and the rumors (err, source code) were true. Bard has grown up into Gemini. Other than it being an obviously better (and more brandable name), Gemini might actually be one of the best ChatGPT alternatives. More on that.

What does it take to be a true alternative to ChatGPT?

1. Comparable LLM (and multimode abilities)
2. Unique value props and features

Is Gemini as Powerful as ChatGPT?

There are others who can better speak to (or have enough time to do so) the technical capabilities of each platform and their underlying foundation models.

You can see countless benchmarks testing each in various subjects (legal, medical, history, etc., etc.). And Gemini is quite comparable to ChatGPT in most of them.

I’ve found that there are a few things that I prefer about Gemini (previously Bard), but they mostly fall under the unique features that Gemini has (assuming parity in foundation models, which is of first-order significance).

How ChatGPT Missed the Mark

When ChatGPT came out with Custom Instructions, I was ecstatic.

After using it since the summer of 2023 (when the feature launched), I’ve become less excited about them. I truly haven't seen a lift in experience when using them (just more likely that I get frustrated when it doesn't follow my instructions).

Same story with Custom GPTs. It seemed very cool and helpful at first, but most of these GPTs are simply a masked prompt over which you have zero control (when using others').

GPTs have largely been a letdown (the API side of GPTs is much more interesting and something I’d like to find more personal usage for).

Interestingly, Gemini has neither of those features, even though it could have easily duplicated the features and released them with the rebrand.

For some, that is a glaring oversight, but Google is a different beast than OpenAI.

Its strategy has to be different because the user base it serves is different (and larger).

As we see the “Gemini Era” unfurling, we can see that its tack is much closer to Microsoft's (unsurprisingly).

What Gemini Gets Right

Gemini is better suited for someone like me. I represent a Google Power User, so maybe that is obvious.

However, it is appreciated when a company does its research on its strata of users.

• I have been a Gmailer for over a decade.
• I use Google Docs whenever I can.
• Google is my search engine.
• I recently signed up for Google One to expand my personal storage (and my family’s).

ChatGPT has ultimately failed to revolutionize any of those things in the year since it has been out.

Smaller entities, on the back of OpenAI’s API, have been more eager to get their Chrome extensions into my hand (largely failed attempts at creating the simplest of application layers overtop of Chrome).

Gemini is simply making its AI more compelling to established users instead of reinventing themselves or copying ChatGPT.

Gemini’s Approach

Firstly, Gemini Advanced is purchased through Google One. This is brilliant for multiple reasons.

1. It extends the usefulness of an important product of Google’s that is a little light on features (Google One primarily offers additional cloud storage with fring benefits of a simple VPN, cloud image editing, more scheduling options in GCal, and access to Google “Experts”).
2. Gemini lifts my whole experience with Google (this is a promise at this point, but Gemini will replace and update Google’s Duet AI) by bringing the advanced AI into places I already go to create (Docs, Gmail, etc., etc.).
3. Cost. I get access to a comparable set of foundation AI models at the same cost as ChatGPT, plus everything I’ve already mentioned ($19.99/mo).

Through all of this, Google gets deeper adoption from their users, and users get AI features (and some fringe benefits) deeper into their natural workflows.

Secondly, this approach, again, is very similar to Mircosoft’s premium AI offering, CoPilot, but it seems better executed.

Everything with Mircosoft has always seemed more burdensome and complicated (for personal users, e.i. normal consumers).

Gemini’s offering seems head over shoulders above CoPilot.

And I think that’ll be even more obvious when Google announces what its plans are for ‘Gemini for Developers’.

Thirdly and finally of note, Gemini isn't blindly assuming that OpenAI is creating the playbook. Google is adapting and doing its own thing.

The people who have said that Google has been playing catch up have only been partially correct.

Google has been playing catch up in as much as they needed a compelling offering that uses the same technology. They did not, however, need to catch up in creating a ChatGPT clone, which is what most of the conversation was about for a year.

Final Thoughts

I actually like how Google is approaching AI more than OpenAI or Microsoft at the moment.

I might make the switch to Gemini Advanced (I’m on a lower Google One tier than what has Gemini Advanced), but first, I’ll start switching over slowly by using and adapting my cache of prompts and workflows to the free version.

If 2023 was the year of AI astonishment, 2024 may be the year of AI usefulness. Competition only makes this more likely the case.

How have you taken to liking Gemini?