Close

ChatGPT Climbs into the Ring with Four Contenders

March 12, 2024

1 min 25 sec read
27 Facebook Twitter Whatsapp Facebook Twitter LinkedIn Pinterest WhatsApp Copy Link Your browser does not support automatic copying, please select and copy the link in the text box, then paste it where you need it.
ChaptGPT has been riding its wave of glory for a while now, and even four weeks ago, it remained the reigning heavyweight champion in the AI boxing ring.

But all good things must end, and now ChatGPT faces challengers for the title.

Virtual AI Fist Punching Through Cyberspace
While ChatGPT-4 has remained at the top of every critical benchmark for some time, the last four weeks have seen the release of four new models to the public that benchmark at or above ChatGPT 4.

The contenders are Google Gemini 1.5, Mistral Large, Claude 3 Opus, and Inflection 2.5.

Each model has its strong points. Google Gemini, for instance, boasts a one million-long token context, which is about eight times the length of GPT-4 Turbo, while Claude seems superior for code. Mistral fares well on devices where others fail, and Inflection's conversation-based interface benchmarks far better than ChatGPT.

Being contenders doesn't mean every one of these models will knock ChatGPT out of the ring. It just means that ChatGPT had no real competition a month ago, and now it does.

Some of the downsides of these contenders include the fact that none are open licenses and no weights are available. After more than a year of huge steps forward in the open-license model category, it's a bit of a letdown to see the best models clinging to the proprietary approach.

And none of these models are transparent about training data, which isn't a big surprise when you consider the lawsuits over training on unlicensed copyrighted data. The public is already a little edgy about the shady nature of AI and how it gathers information.

Transparency is one of the biggest issues with the contenders, and while training on only public domain and licensed content might be the admirable thing to do, it's likely this isn't happening.

Knowing how a model is trained is a big factor in knowing its strengths and weaknesses, and in the absence of transparency, we're all left with guessing games.

One thing is clear. The battle for AI supremacy is raging, and while ChatGPT faces some heavy contenders, it's not even close to down for the count.

Want to read this in Spanish? Spanish Version >>

Categories: Apps, Technology
27 Facebook Twitter Whatsapp Facebook Twitter LinkedIn Pinterest WhatsApp Copy Link Your browser does not support automatic copying, please select and copy the link in the text box, then paste it where you need it.