Google recently unveiled its riskiest bet in the field of generative AI, a bet that attacks head-on the dominance of Open AI and its ChatGPT: Gemini is its most ambitious model, a model suitable for any device, including Android. And now you can try it out: Gemini Pro is active on Google Bard if you follow a very specific process. We’ve given it a try.

The dominance of ChatGPT was making Google nervous, not in vain Open AI overtook the search engine company by taking advantage of its training and machine learning tools. It was time to turn the tables, Gemini is the multimodal model that comes to dominate them all. In fact, there are three models: Nano, intended for devices such as Android; Pro, for human interaction tools; and the Ultra, an advanced model with which Google intends to revolutionize from the research to development segments. And Gemini Pro can now be tested.

Google Bard Finally Lives Up to What Google Deserved

I’ve been testing Bard since Google launched it in private betas, and the truth is that until now it hadn’t evolved too much: I recently compared it to ChatGPT and it kept failing miserably. The AI didn’t reason creatively enough, the data often wasn’t correct, I didn’t find the code developments satisfactory either. So, with the arrival of Gemini Pro, it became clear to me that I needed to put it through its paces to see if the new model had solved the shortcomings of its predecessor.

For starters, it’s not too easy to test the new model because it’s only available in English. However, it is a matter of applying a few steps and knowing how to get along with the language:

  • Sign in to your Google Account and scroll down to “Personal Info.”
  • Scroll down to “General Preferences for the Web” and go to “Language.”
  • Select U.S. English as your preferred language. Alternatively, you can go to this link by avoiding the steps above.
  • Now you need a VPN to simulate a connection from the United States. I chose NordVPN.
  • Load a US server and go to the Bard page. It should come up in English letting you know that you’re using the Gemini Pro model.

Once I had Google Bard active in my Android web browser, and after making sure that Gemini Pro was working, I asked it a number of questions related to different areas. Direct questions about historical data, mathematical calculations, problem-solving, or creative reasoning around philosophical questions. I found a clear evolution in behavior, both in terms of satisfactory results and in the way of reasoning. Although I saw some difficulties when it came to following the thread of the conversation when I returned to a concept already discussed, something that does not happen with ChatGPT.

Google claims, precisely, that Gemini Pro is superior to GPT-3.5, at least at the level of neural performance tests. And the truth is that they are at a very similar level, at least according to the tests I have done. Of course, against GPT-4 Google Bard has nothing to do, at least until Gemini Ultra and its multimodal operation are available; something that will happen at the beginning of 2024.

I really miss being able to ask for images, for example, also a more creative reasoning: Gemini Pro is a bit cold, it has a hard time getting out of logic; even though he surprised me in more than one response, such as when I asked him for a Christmas card and he wrote it to me in Spanish because he sensed that I was in Spain (despite the fact that Nord VPN was connected to the United States). It shows itself faster than its Open AI opponent, the response speed is at an excellent level.

At the moment in tests restricted to the United States

I’ve been testing the new Bard, both from the smartphone and on the computer, and the result was satisfactory in both for all tests: where Bard was shipwrecked before it is now much more reliable. It is not the best way to consult doubts, but it is gradually getting closer to what a generative AI should be at the query level. Multimedia generation would be missing.

Greater reliability, better reasoned answers, increased breadth of action, remarkable speed and even a certain approach to human logical thinking, always bridging the gaps posed by a Training-based Artificial Intelligence: Google is doing its homework, even if it is (much) late. I’m looking forward to seeing what Gemini Ultra has to say, we’ll see it at the beginning of the year with an experimental version for Google Bard.

Currently Gemini Pro can be tested, just change some Google account settings and use tools like Nord VPN.

Comments

LEAVE A REPLY

Please enter your comment!
Please enter your name here