site: gemini.google.com/
Google is in an extremely precarious position as the AI tsunami approaches. On the one hand it is one of the most technically able of the competing companies. That should put it in a position to produce the best possible tools in all areas of AI. On the other hand it earns a great deal of money with 'search'. Indeed the word 'google' is synonymous with seach. AI conversational tools can often solve problems without the user needing to do any intermediate research - so no 'search' stage at all and hence no potential for advertising revenue.
Google Gemini at March 2024
Google's Gemini tool combines:
- an AI conversational 'app'
- Search (though of course AI is also heavily used to make that work efficiently)
- an AI image generation feature (apparently built on top of an AI model called Imagen 2)
Home page of Google Gemini. This is very simple.
At the time of writing the image generation is 'paused' but will doubtless soon be back. See the link on our home page for at least Google's take on the very interesting reasons behind that.
Disclaimer. It is worth reading the Google Gemini terms/disclaimer which states amongst other things the key thing about these models in their current state of development - they may be inaccurate and they may give offensive answers.
Prompt entry. The Gemini basic 'prompt' screen is like many others. You can enter anything you want and there are a series of suggestions for those who do not know where to start. See also the opportunity to speak your prompt or even to upload a supporting image.
Note also the line at the very bottom about 'inaccurate information about people'. That is very common which is why the AI Libel site exists. Having a small note at the bottom of the screen may not be enough to protect against a defamation prosecution in the UK or elsewhere.
Answer. The Gemini answer is, as usual for these tools, cautious but the ideas are very much in line with what we believe here.