Brought to you by
Aquilla_highres

A Cloud-based ERP solution for mid-market organisations

Save content
Have you found this content useful? Use the button above to save it to your profile.

ChatGPT: This changes everything!

25th Jan 2023
Brought to you by
Aquilla_highres

A Cloud-based ERP solution for mid-market organisations

Save content
Have you found this content useful? Use the button above to save it to your profile.

ChatGPT: This changes everything!

 

The last couple of years has seen dramatic development in AI (Artificial Intelligence). Every time a computer needs to recognise text (OCR), voice or images it is done via a trained neural network. Text and voice recognition are practical at a human level. However, image recognition, at least in specialised cases such as cancer detection, far surpasses human expert capability. On the other side of this is generative AI, which is far more complex since it involves creativity – something that humans often like to say will never be the domain of computers… well, “hold my beer”, says OpenAI! In this article, we’ll be discussing the latest advancement in ‘creative’ AI: ChatGPT, its limitations, dangers, controversies and its significance for the future of finance.

 

How does it work?

 

Imagine training a neural network to recognise a dog by showing it numerous pictures of dogs. The AI itself will use this to decide the distinguishing features of the dog and use it to identify the specific animal in the future. Now imagine you train the same AI with every piece of information and how it is all connected: every book, image, video, website, etc. OpenAI did just this and created a model (of the world) called GPT-3. GPT-3 ‘understands’, or recognises, the world like a human that is trained, to college degree level, in every field of expertise that exists. As AI is so good at extracting patterns, it ‘knows’ the style of Van Gogh versus Picasso; Shakespeare versus Eminem; or Pong versus Call of Duty. AI is able to infer meaning and intent from vague requests and questions, even self-correcting when an error is indicated.

 

This latest development was, for many, considered the stuff of science fiction, alongside cold fusion or even faster-than-light travel! It’s no surprise that this new AI phenomenon has been branded as ‘revolutionary’, it is of course essentially a conversational, ‘Star Trek’ computer interface! Thankfully, however, it is not sentient or alive, nor does it have wants, needs, will or its own agenda. In fact, it’s curated to barely form an opinion about anything. So far it is a tool… a very, very capable tool. However, it is built and trained solely on human knowledge, which is incomplete and prone to error, meaning it can and does make mistakes, although probably less than the average human tasked with the same request...

 

Read the full article here