TESTED: Google Gemini Pro vs. GPT-4 (Bard vs. Bing Chat Copilot)

Real world testing of Gemini Pro vs. GPT-4 on various topics.
Google's Bard is using Gemini Pro (previously it used PaLM2).
Microsoft Bing Copilot is using GPT-4.
On about half of them the two language models are about equal.
On the other half GPT-4 is better, particularly for the python code.
In the pharma literature search the difference was significant.
At the time of upload (07Dec23) Bard is not using Gemini pro in most European countries.

0:00 Intro
0:12 Sport Trivia
0:53 Investing
2:19 Distance Calculation
4:23 Science Question
5:26 Python Code
6:29 Math Proof
7:00 Pharma Knowledge
9:47 Chuck Norris Jokes
13:15 Summary

Bard gets Gemini:

Videos:
Main google (good for Broll):

Hands on (from Google):

Programming:

Genini Availability:

Blog Post:

Report:

[ForMyDawg]

Bing chat and gpt-4 are very different. The real gpt-4 is significantly more advanced than any mode of bing chat. It’s inaccurate to refer to bing chat as gpt-4 (even though it is partially powered by gpt-4).

[The-Rest-of-Us]

Tip: on Bard you can press the little “G” button at the bottom, then it fact-checks its answer and provides sources. Great function, works so so most of the time.
Also side note: Bard runs Gemini Pro, which should be compared to GPT 3.5.

[mesh]

What you were using is the creative version of bing chat, it tends to hallucinate more as it is geared more towards generating poems/stories, not facts.

[luxushauseragency]

Well done. Great questions and conclusion. Very useful.

[carlosforero]

thank you, I was looking for this comparison. Great work

[ilyassffyt]

Good comparison, we are eagerly awaiting for Gemini ultra✨

[pilotgfx]

i think you could enhance the questions you use to give you a better foundation for evaluating, and also which would give responses that are less subject to your personal assessment (less wether u like mutual funds etc but rather just true or false, predetermined criteria to pass)

[fjlgpgd]

in my opinion Gemini answers are more straightforwardu and less time wasting and more clear to read than gpt-4

[garethbaus]

Manhattan Kansas is actually a fairly major city for it's region.

[sammikumar]

Bard often provides incorrect information and is not reliable, whereas you can verify the source in Copilot.

[serial_box]

Are you sure you already have Bard running with Gemini? I have 2 accounts and only on one of them do I get the message that Bard has been updated with Geminini. This is a blue message at the top. On my other account it only says that privacy has been improved, just an older message

[HuGiv]

Amazing video but don't forget that GPT4 competitor is Gemini Ultra... This should be compared to GPT3.5...

[WMCheerman]

I have been finding that Bard is excellent at sifting through academic papers. I have found that prompting bard that looking through scholarly papers is my intent first helps. I assume it due to google scholar.

[allanomondi]

Anyone noticed some bias against Bard?

[benyamin]

chat gpt 4 is 20 dollar per month compare to free gemini, so that would be a huge win for gemini...

[idkidkdidid]

i think bard really did a great job here

[gutlibe]

Nice video 📷

But not that bing chat is optimized to generate a concised output. So you can try comparing it with the corresponding one for better

[Samtagri]

This is not gpt 4. This is bing search which is much shittier than gpt 4

[mabonhunts]

In your title you put Genini but your are actually using Bard I would like to see you re-do this test with GEMINI vs Copilot instead of Bard. p.s. next time don't lie in your title.

[V-obzf]

To be honest you were pretty bias while comparing them. You picked a side. Do you agree ?