ChatGPT v Bard

[Snyper2020]

I've recently started using GPT, I find it reasonable useful but extremely biased, quite frustratingly.

I started using Bard more recently and while I believe they operate differently I've found Bard more unbiased.

The acid test is asking something like, Is climate change a hoax, and while I think most accept its isn't both GPT and bard will support it is not a hoax. However if you ask both for any evidence to the contrary, GPT won't answer, Bard will however, albeit you are likely to dismiss or not agree, it gives information.

 

GPT is much more likely to give an "i cant answer that" type responses where Bard will give both sides

 

I've found Bard so far better to use and with the integration eventually of Google word and sheets it will be better 

 

Bard V GPT 

 

The iPhone v Anroid debate for the next decade 😆 

 

[pmrd]

Let me guess, ChatGPT is too woke?

[Good Bot, Bad Bot]

  On 22/11/2023 at 17:12, PmRd said:

Let me guess, ChatGPT is too woke?

Expand  

Hey I need my AI to tell me maybe the last election was rigged. You know to make sure it's "unbiased".

[The Rev]

  On 22/11/2023 at 17:21, Good Bot, Bad Bot said:

Hey I need my AI to tell me maybe the last election was rigged. You know to make sure it's "unbiased".

Expand  

Please tell me you're joking.

[deutscheclub]

i think they both are good they helps alot in your projects and other stuff

[Good Bot, Bad Bot]

  On 22/11/2023 at 17:41, The Rev said:

Please tell me you're joking.

Expand  

Yes, I didn't think that needed to be said.

[Snyper2020]

  On 22/11/2023 at 17:12, PmRd said:

Let me guess, ChatGPT is too woke?

Expand  

I think you're missing the point

Edited November 23, 2023 by Snyper2020

[AStaUK]

I've not used Bard much, I've done a little more with GPT. With GPT I've used it to produce simple powershell scripts, or fix something in a script and for the most part that works really well, where I've found Bard will output a script but if you ask it to make changes even simple ones it quite often doesn't understand what you're asking. With GPT you need to be wary of the responses, it will respond using the data it's been trained on, this was up to Sept 2021, it may have been updated since. It's also been reported that GPT can "hallucinate" and make up an answer where it doesn't know one, I suspect this is true of other AI's as well.

But it's early days for this technology and businesses and the general public are still working out how/where to use it and the impact it will eventually have on our lives and over the coming years the technology will improve drastically.

[Snyper2020]

  On 23/11/2023 at 10:48, AStaUK said:

I've not used Bard much, I've done a little more with GPT. With GPT I've used it to produce simple powershell scripts, or fix something in a script and for the most part that works really well, where I've found Bard will output a script but if you ask it to make changes even simple ones it quite often doesn't understand what you're asking. With GPT you need to be wary of the responses, it will respond using the data it's been trained on, this was up to Sept 2021, it may have been updated since. It's also been reported that GPT can "hallucinate" and make up an answer where it doesn't know one, I suspect this is true of other AI's as well.

But it's early days for this technology and businesses and the general public are still working out how/where to use it and the impact it will eventually have on our lives and over the coming years the technology will improve drastically.

Expand  

I asked Bard a question I knew the answer to yesterday regarding an engineering process.

 

The answer looked great, but mixed up and intertwined the procedures from 2 completely different products and process from different continent's!

I've found with GPT is can get the most basic math problems incorrect, so I have a habits if asking it to recheck its answers to which jt will usually more often than not get it right.

 

Other occasions it will insist it's right, and only when I point out a flaw it will "apologise" and recalculate or evaluate 

 

I find AI useful and see great potential but it seems its only as useful as the information it draws 

[AStaUK]

AI is no different to any other application, garbage in garbage out. When setting up services using AI the governance is as important as the technical side, you need to validate the information you are training the model on is accurate and you need to secure the training material and the model itself against tampering from either curious individuals or those with malicious intent.

AI has been around for a little while, what has changed is that it's now more accessible and we're now starting to see applications emerge that are making use of it and we're only at the start of that journey and have a long way to go. For now if you're using AI you should still fact check the output and not take it for gospel.