Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Google Opens Up Access to Bard AI Chatbot

Is there a 'clean internet' I can use that doesn't syphon everything I do, say and write on the internet?

I know, I know: "if you don't like it get off the internet" etc etc.

But it would be nice. I use MS Edge on my Mac and now there is that persistent Bing button, and its just letting me know its sucking down everything I do for the sake of mankind.

Screeshot 22-03-2023-05-22.jpg
 
ChatGPT seems to have a strong inherent bias about many topics. Those biases will become a major problem if it turns out that really smart AI needs such a huge number of computers that only large corporations can afford that. My dream was a smart computer that could sit on your desk and is completely independent from Google or Microsoft. The problem is that the centralized AI with a million computers and trillions of parameters will always be much smarter than the one powered by a single computer on my desk. That will likely still be true in 100 years from now. As A result, AIs will make those coprporations even more powerful.

For example if you ask ChatGPT to tell you any advantages of climate change, it will refuse to do that and instead tell you a lot of disadvantages of climate change. That was not the question though. Even if you think that climate change is really bad, you might want to know what arguments people with a different opinion might come up with. For example less people freezing to death in winter or stuff like that. ChatGPT seems to be programmed to try avoiding any answers that could damage its reputation. You could try the same with other delicate topics like gun violence.

All those AI companies still remember the disaster when that one AI (was is one from Microsoft?) was turned into a racist by trolls who had "fed" it with a lot of racist text. So they are now extra cautious to avoid any bad press. That spoils some of the fun with AI.
 
  • Wow
Reactions: gusmula
SBlue1 said:
I just had an hour long chat about physics inside stars with ChatGPT and I asked a few times for sources to double check its answers and sources seemed to be credible to me. It would be great if it would have access to recent scientific papers as well. Or imagine it having the all the microfilms of newspapers from the beginning of the 20th century so you could ask what the weather was like in New York when the Hindenburg Zeppelin exploded or what was the top 10 headlines in newspapers the day the Titanic sank.

Didn't some museum collect tens of thousands of letters soldiers sent home during WW1, so imagine asking what did the regular person think of the war in the first weeks in comparison to the last year.
Click to expand...
I'm not sure how my experience will compare to yours, but I've had similar academic discussions with ChatGPT, and when I've asked it to provide references as a follow-on question it will happily do so - but the references invariably seem to be incorrect. The source publication is always appropriate to the topic of discussion, but when I drill down to the referenced article (e.g. by doi or page/article/edition etc), it's always a different article by different authors - usually on a similar topic, but it's not the source of what ChatGPT's said, nor does it generally provide evidence to support it. I did wonder if it was finding articles based on their references, but never did find any connection between the content and the reference ChatGPT provided.
 
klasma said:
From what I’ve seen so far, it seems to be rather poor in comparison to ChatGPT.
Click to expand...
My experience exactly. I’ve done about ten different comparisons with the same prompts. Bard could not write a Bash script but ChatGPT produced one. I tried having both write short stories. ChatGPT was much better and coherent. Bars did okay with a couple questions about animals but ChatGPT did equally well on those.

Overall, I’m very unimpressed with Bard. It will improve but it’s much worse than ChatGPT was at release.
 
  • Like
Reactions: BigMcGuire
Can someone make a script that will copy paste outputs from one bot to the two other’s input and vice versa so they can engage in an endless conversation and leave the rest of the world alone?
 
NightFox said:
I'm not sure how my experience will compare to yours, but I've had similar academic discussions with ChatGPT, and when I've asked it to provide references as a follow-on question it will happily do so - but the references invariably seem to be incorrect. The source publication is always appropriate to the topic of discussion, but when I drill down to the referenced article (e.g. by doi or page/article/edition etc), it's always a different article by different authors - usually on a similar topic, but it's not the source of what ChatGPT's said, nor does it generally provide evidence to support it. I did wonder if it was finding articles based on their references, but never did find any connection between the content and the reference ChatGPT provided.
Click to expand...
It makes up references. It’s a generative language model and it will generate answers that could be correct but are not.

I asked ChatGPT to list some of my published research articles and it made up some reasonable sounding ones with people I could conceivably have published with but none of the references were real.

ChatGPT can produce excellent and accurate information but it will confabulate information (others use the term hallucinate but that’s not what’s happening, it’s a confabulation).
 
  • Like
Reactions: aleksander and NightFox
mystery hill said:
Bard. Who comes up with these names?
Click to expand...
Im sure the name is Googles 'Hilarious' nod to William Shakespeare, being a prolific writer and inventing many words that we use in modern language.

Bard being a language learning model and providing long form answers. But that's just a guess.
 
ninecows said:
Can someone make a script that will copy paste outputs from one bot to the two other’s input and vice versa so they can engage in an endless conversation and leave the rest of the world alone?
Click to expand...
That reminds me of an prank they did in a German TV show. They called a sex hotline and and astrology hotline at the same time and then connected those two to see what would happen. I don't really remember how it went, but it was funny.
 
  • Haha
  • Like
Reactions: ninecows and wilhoitm
So my guesses are:

50% chance they named it after the jack-of-all trades Bard class in Dungeons & Dragons.
50% chance they named it after the character Bard from The Hobbit who slayed the dragon OpenAI Smaug.

But I'm 100% sure someone at Google has terrible naming sense. :p
 
Kind of serious question: should we trust Microsoft or Google more with AI inquiries tied to our online information/identities?
 
I signed up and got access to Bard in less than an hour.

I was generally unimpressed. Whenever I asked it a question that required some sort of critical thinking, I got an answer back that sounded like it had been written by a lawyer in corporate.

Bard, will my cat be hungry when I get home?

"If your cat hasn't eaten for most of the day, your cat may be hungry. However, some cats are known to be able to survive for quite some time with little food. You cat may or may not be one of these cats. If you feed your cat when you get home, he may or may not still be hungry.

It was more like a weasel word generator that went to a tier 3 law school in Delaware 🙄
 
  • Haha
Reactions: compwiz1202 and JM
haruhiko said:
Meet Bard: your creative and helpful collaborator, here to supercharge your imagination, boost your productivity, and bring your ideas to life.

Bard isn’t currently supported in your country. Stay tuned!
Click to expand...
Google must not think your country has any good ideas worth bringing to life. What a terrible company 🤣
 
tothemoonsands said:
We are on the precipice of a major technological leap. Specialized AIs will soon emerge and then eventually consolidate. The improvements with each iteration are going to rapidly accelerate. I’m very optimistic with what this could do to world’s GDP, but also efficiency in terms of environmental sustainability. There will surely be some dystopian effects of this technology, but the good outweighs the bad.
Click to expand...
AI still can’t do abstract reasoning. It’s still based on statistical analysis from very large data pools. Abstraction is the next technological leap.
 
  • Like
Reactions: tothemoonsands
Feels like Google is trying to push Microsoft downward in the AI department... considering Google's Pixel lineup has been a solid winner in smarts and Microsoft's last attempt for a phone blew up (not in a good way). So far, I still think Microsoft is winning... I might try it out later but I'm not expecting much.
 
Michael Scrip said:
I wonder if AI chatbot technology will be at the level of importance as something like the smartphone?

Or will it be the next 3D TV ?

🤔
Click to expand...
I don’t think anyone at this point seriously entertains the idea that AI chatbots are a temporary fad. 3D TVs never were as popular or used (as 3D televisions) as much as these generative AIs already have been. The number and diversity of potential (and current) applications is so big that it is practically impossible these won’t keep growing and become an important life of our lives.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back