Google Bard: Is It Better Than ChatGPT?

📝 The paper "PaLM 2 Technical Report" is available here:

My latest paper on simulations that look almost like reality is available for free here:

Or this is the orig. Nature Physics link with clickable citations:

🙏 We would like to thank our generous Patreon supporters who make Two Minute Papers possible:
Aleksandr Mashrabov, Alex Balfanz, Alex Haro, Andrew Melnychuk, Benji Rabhan, Bryan Learn, B Shang, Christian Ahlin, Eric Martel, Geronimo Moralez, Gordon Child, Jace O'Brien, Jack Lukic, John Le, Jonathan, Kenneth Davis, Klaus Busse, Kyle Davis, Lorin Atzberger, Lukas Biewald, Martin, Matthew Valle, Michael Albrecht, Michael Tedder, Nikhil Velpanur, Owen Campbell-Moore, Owen Skarpness, Rajarshi Nigam, Ramsey Elbasheer, Richard Sundvall, Steef, Taras Bobrovytsky, Ted Johnson, Thomas Krcmar, Timothy Sum Hon Mun, Torsten Reil, Tybie Fitzhugh, Ueli Gallizzi.

Károly Zsolnai-Fehér's links:

#chatgpt #bard

[_koy_]

great to see ai companies competing. that will make ai 100X better

[rijaja]

I wonder what kind of privacy violation loopholes they still need to go through to be able to release this in the EU

[DangerDurians]

We are rapidly approaching a point where Cortana might be a useful windows feature

[dharveyftw]

I found bard really great for helping me do research for my papers, since its able to brows the web, but using GPT to refine the information I have, using these different LLMs together is the way to go

[karlvuleta]

I use GPT-4 for coding with C++ as I'm not familiar enough with the language to code with it on my own, and GPT-4 is absolutely fantastic at interpreting instructions in plain English and producing code that works nearly always the first time. In comparison Bard feels unusable at times, and regularly misinterprets instructions even when broken down in technical terms.

Bard on it's own is extremely impressive generally, however I have been absolutely spoilt with GPT-4 that Bard fails to meet my new heightened expectations... for now

[ArielTavori]

I tried almost the exact same request asking bard to list five art colleges in a certain area, and then in the next prompt asked it to show those on a map. It said "sure, here they are marked with pins on a map" and then showed me a picture of a physical map that I could buy from some store. Epic fail.

[sanketvaria]

so bard is improving. good, now it's time for open AI to release their locked features into GPT4 for public.

[markemad]

Let's not forget how chat gpt got at least 2x worse technically not to mention practically after the security implementations, we absolutely don't have any access to the state of the art sadly even the one we had just a bit ago.

[maninalift]

So far my limited experience with bard is that it gives answers that are more systematic and explains it's raining more but is also less creative and generates less interesting ideas.

[justinlloyd]

Bard tends to hallucinate when coding. Oftentimes the code won't even run. Bing tends to always run. Bard does fine in other tasks though.

[DeFaulty]

I would argue that the success of ChatGPT is owed more to cooperation than competition.

[KamilBanc]

While it's an interesting war to watch, I'm more hopeful that decentralized business models like stability AI will take the rain, and eventually lead the crowd because I do think that AI will become as essential as other human rights and because of that, I would not want it to be regulated by a few giant companies, without any viable alternatives for the individual.

[vslaykovsky]

There are no more "papers" from big tech companies. They are replaced by "tech reports" which are quite useless for research purposes.

[welltell.]

Well there is one error in the pictures!! Where are the clocks??? or Timers???

[CaseyConnor]

But how is the accuracy? These bots so far have not been useful to be because of the confabulation problem... I end up having to spend as much time confirming their information as if I had just found the answers with conventional methods. I'm sure it will get much better quickly, but are we there yet?

[Cingku]

While Bard currently falls behind ChatGPT in terms of reasoning (although I believe it has the potential to catch up in the future), it surpasses ChatGPT in knowledge. Google, being a search engine, indexed the entirety of the Internet, granting it access to vast amounts of information. On the other hand, ChatGPT's knowledge is limited, and despite utilizing plugins to browse the internet, it often fails due to its bot nature, as websites tend to disfavor interactions with bots. However, once Google manages to achieve parity with GPT-4 in terms of reasoning abilities, it will undoubtedly become the most formidable chatbot in existence.

[jeromej.]

Bard is not available in your country right now/yet.

But chatgpt is. :(

Virtual borders suck and I can't believe we still have them in 2023.

[skylineuk]

Gpt4 is still miles ahead. I’ve been using both since their betas came out and I got early access and Bard is still unfortunately well behind in this race.

[duudleDreamz]

In my experience GPT4 is far superior to Bard. I test Bard every few days to see if it has caught up, but quickly return back to GPT4. GPT4 + Code Interpreter is incredible. I don't think this video makes a fair comparison.

[frankgreco]

I typed those prompts into DallE-2, and I didn't get anything close in quality to what you got.