A lot has been spoken about artificial intelligence (AI) by people far wiser than I, but I wanted to focus on a specific part of the story that I became aware of, following a Radio 4 interview between Nick Robinson of the Today programme and the CEO of Mumsnet.
The news item they were discussing was the proposed legal action being taken by Mumsnet against Open AI for the alleged unauthorised use of their intellectual property, namely the content of the Mumsnet web platform. Now, to fully appreciate the impact of this item, you need to understand a little bit of how AI such as Chat GPT, Co-Piot etc work.
AI as we commonly experience it is actually a series of programmes called Large Language Models (LLM). LLMs effectively spend all day scouring the world’s web pages for words, sentences, questions etc – basically, language in all its forms and structure. This means that when a user types in a question to an AI programme, the LLM will cross reference all of the content it has examined to provide what it predicts is the most suitable answer to your question. All this, in the blink of an eye. Very impressive! And probably why LLMs need so much power to operate.
But, and here’s the rub, the answer provided by AI was derived from your content. Did you give AI permission to use your content? Did you know that LLMs will “scrape” your entire web platform, including chat rooms, blogs, page content and resources, regularly? And this was the point that Mumsnet was making. LLM providers should not have unfettered access to all code on the internet. There should be a value exchange and a recognition of intellectual property considerations.
A business is more than happy to allow Google to scour its web page content in return for being listed and found by potential customers – this seems a fair value exchange. But what if you provide legal advice to a client’s question on your platform as part of that web service? Are you happy for your answers to be used to answer a question from someone else with no fee or acknowledgement?
Whatever, your view on the benefits and disbenefits of LLMs, I suspect we are all about to become acutely aware of who has access to our data and what they’re allowed to do with it. For decades, we have been reminded that we all put too much data about ourselves onto the internet and allow unknown agents unrestrained access to our apps. GDPR was an attempt to manage concerns for the consumer, but it seems no one was paying attention to the back door. I see troubling times ahead on this one.
The overriding message that we all came away with was the need to address how our businesses will be affected by AI. Not one speaker left the stage without commenting on this so we will be looking at how we can support businesses in this space going forward.
