Google lately rebranded its AI chatbot Bard to Gemini, and with it got here a brand new paid tier known as Gemini Advanced, priced at $20 per thirty days. This improve guarantees entry to the newest and most superior Gemini Ultra 1.0 mannequin.
The tech big didn’t maintain again from telling everybody how nice the Gemini Ultra was speculated to be. The manner they talked about its skills was actually spectacular, and I used to be satisfied it was going to be a giant step ahead in AI.
However, after testing it out for a number of days, I’ve seen some points that I feel folks ought to learn about earlier than switching to the paid model.
- Gemini Advance’s response may be very sluggish
- It sometimes fails at fundamental logic
- Some drafts are incomplete
- Image era bias
- Weaker picture understanding than ChatGPT
- Gemini Advanced can not zip or obtain recordsdata
Let me clarify every of those points in additional element.
1. The response may be very sluggish
Gemini Advanced disappointingly lags behind in response occasions, taking roughly 5 to 7 seconds, in comparison with ChatGPT’s 2 to three seconds.
Check out the side-by-side velocity comparability of the 2:
This delay is constant, even when accessing extra drafts, which equally take a very long time to load (round 5 to 7 seconds).
2. It fails at fundamental logic
Gemini Ultra struggles with fundamental logical reasoning. For occasion, when prompted with a easy mathematical query relating to automotive possession, the AI clearly fumbles.
Prompt: Today I personal 3 vehicles and offered 2 vehicles final yr. How many vehicles do I personal?
If you ask the identical query to ChatGPT utilizing the GPT-4 mannequin, it will probably reply the query…