Earlier this month the AI world watched in awe as Google launched Gemini, its Chat GPT killer.
A sexy demo showcasing Gemini’s ability to respond to live video feedback, and stats demonstrating it outperform GPT-4 (Open AI’s latest AI model) suggested that Google was now the king of AI.
But when Gemini got in the hands of users, many concluded the demo was staged, and that its results were underwhelming. This led to the question: is Gemini actually better than Chat GPT?
The answer may actually lie in a game: enter the Chatbot Arena.
Chatbot Arena allows users to ask the same question to two AI models at once, with the winner providing the best answer. Watch this video as I ask a random question that requires unique of logical reasoning: How many tires do you think are in the city of Detroit, Michigan? I can continue my chat battle until I’m satisfied with the best answer, and Chatbot Arena then reveals which model I used.
Stack this up hundreds of thousands of times and you have a very convincing leaderboard on which model is the best.
Chatbot Arena recently updated its leaderboard, which now includes Gemini, and the rankings are eye opening:
-Open AI’s GPT-4 is the undisputed winner with three of its models (the original launched in March and updates in June and November) holding the three highest positions
-Google’s Gemini ranks an unfortunate 12th
-Mixtral, a free model founded by Google Alumni outperforms Gemini in 8th place
If the Chatbot Arena rankings have anything to say about it, Google remains far behind in the AI arms race. What more, free solutions look to make powerful AI systems free and accessible to anyone.
