Artificial Intelligence is featuring heavily in the news. Firstly, this is due to the unprecedented level of investment in computing capacity by the AI companies. This investment is in processing chips especially from companies like Nvidia, and in the data centres necessary to handle the computing and storage.
But secondly, there has been growing concern that AI will do irreparable harm, perhaps a threat equal to the Cold War.
We gathered at The Witham to learn more about this new technology. Pete Redgrave talked us through what AI really is, what it can be used for in everyday life, and what the risk around its development might be.
AI has many uses: Robotics; Face recognition; eCommerce; Data security; Digital assistants; Navigation and Education to name a few. The workshop concentrated on chatbots which are readily available to all computer users.
A chatbot is a computer program designed to simulate human conversation through text or voice. Common examples are ChatGPT, Perplexity, Claude and Google Gemini.
Users interact with chatbot, behind which is huge computing power, known as a Large Language Model. These LLMs have learnt how to interact with users by hoovering up vast quantities of data. Trillions of pieces of information including practically every book ever written.
We learnt how to interact with a chatbot, especially how to frame questions. This framing shapes the answer for example the level of detail, the format and constraints.
Chatbots can reply at any desired level, from bedtime story for a toddler to a doctorate level thesis. They can answer factual questions, make recommendations and provide technical support. They can produce images such as this one that was generated by requesting a donkey playing basketball on Bondi Beach.
AI is unavoidable. It is in the background of many online transactions, generating correspondence and creating transactions. It has massive potential in areas such as medical research, design and engineering. But there are significant ethical and safety concerns.
Chatbots are programmed to avoid harmful topics; however, they do have negative aspects. These include but are not limited to inaccuracies and “hallucinations”, over-reliance and skill erosion, bias, security and misuse.
At the end of the workshop, we left better informed and perhaps more confident that using chatbots would not herald digital Armageddon!
(This article was written by a real u3a member not a chatbot.)
Report by Peter Colley