Google’s Flubs Bard

[pepsiicat]

how do you lose 120 billion from a failed launch of your own bot you designed💀

[Withered]

OpenAI is so suggestive because it's designed to be passive and polite. After a certain point it's not worth arguing with the user and it'll just agree with them.

[wasdc]

ChatGPT isn’t any better. I was asking it some pretty advanced chemistry calculation questions, and it failed to get a single question right. So I tried simpler question, such as how many litres of gas in one mol at STP, it gave me 25.8L. That is literally on every front page of Wikipedia and it manages to get it wrong. I tried other simple mol conversion questions, I was greeted with fail after fail.

[fall]

I always tell them "that's ok buddy, we'll search together".

[CristianTheRich]

I feel like them pulling the stream and such may have worsened the issue, everyone knows the bots aren't perfect but thats pretty funny nonetheless.

[Akimakida]

Google has the teachable machine thing or something like that and it's pretty cool

[JP_RS]

Im worried how much we are developing AI at all...

[PickledSoupz]

I played chess with it in notation so I just did e4# which would be mate in 1 (not possible) and it was like good game!

[ecologist]

The problem for google is that they have been touting that they are SOOO far beyond OpenAI, and said Bard is a pared down model of Lambda. However, they put up something that is questionable if it's on parity or worse than ChatGPT. They did not come out a mile ahead.

[ZenithEra-]

they made a chess bot that can beat STOCKFISH (3800 ELO)

[MrHendrikje]

The reason ChatGPT 'Corrects" it self if you insists is because is made by OpenAI which is founded by Elon Musk. Elon is VERY wary of any thing AI and be firmly believes AI to be very dangerous but he also acknowledges that it can be a great tool, but ONLY if humans can steer AI and correct it when needed.

So having users insists its wrong, and ChatGPT responding to this is likely a feature and not a bug.

[desi_bhai_]

Microsoft has given the promising note on it

[uditraj]

This was the reason why google didn’t use ai up until now when there is a real competitor as being a big company handicap from doing anything out of the box..

[jarodjagges]

So...AI is the same everywhere. It can't usually distinguish real info from fake info. You can ask one to tell you about a scientific subject as an example, but it will likely use nonexistent references and only half-way succeed. One of my professors tested it out himself and knew instantly because of how small the community around the study of cognition is and he has read many of the important papers it could have used in the first place.

[for]

Chatgpt just knows you can't win an argument against a woman

[iphonehome]

Hear me out! What if the AI isn't wrong, it is just lying to us?

[vspatmx]

Yeah...
I am betting on
Alphabet Google

[michealzachary]

I don't trust them at all the only thing I am looking forward to is that ai enabling the creation of basic dialogue with none important characters without needed to dedicate a writer to each one. For modders and small time video game devs

[carlsonbench]

Completely.

I believe that what we need more than anything else is an uncensored open Ai.

Something that can unequivocally tell use the truth. I want both more than for an uncensored AI to reveal the truth about the events of 2019

Was it deadly?

Was it safe and effective?

[The-BlueHatMan]

These machine learning trends need to die already