For some of us less tech capable people, the sudden introduction of AI technology has us a little overwhelmed. Let’s be honest, Skynet and it’s unleashing of Terminators on John Connor probably didn’t paint AI as a positive force for good.  For those not scarred by Arnie melting in the fiery pits of the steelworks, readily accessible AI has been an absolute boon in terms of productivity and has already changed the way we work.

What are we using AI to do?

The purpose of generative artificial intelligence (AI) technology is to produce various types of content such as text, images, audio, music and so much more. You can use it to write a blog (not this one, it was completely written by a real live person!), prepare a to do list, create a photographic image, compose a piece of music or write computer code all with a well written prompt that directs the algorithm to generate the content you need.  In fact, part of the Hollywood writer’s strike is how AI will be used to displace both creators of content and actors in the future.  Wow! 

It's been around for longer than you think

It does feel like AI has come out of nowhere.  I think we knew it was being developed but it was one of those things that feels a little sci-fi and therefore unlikely to impact our lives much in the immediate future.  In fact, we have been using AI without it calling it that for quite some time. Google Maps and Online chat anyone?

An important thing to understand is that generative AI is created to perform the one specific task it is directed to do, and that is all it does.  For example, it may be used to create articles, or it may be trained to make viewing recommendations (as per Netflix) – generative AI stays in it’s lane and doesn’t cross over to other functions.   

Acceptable usage

Clearly, it’s here in a new format which is easily accessible to assist people with many and varied tasks, and it’s moving fast so it’s really important that you address how it’s used within your business now.  Of course, the easiest way to start is to have a clear policy in place that establishes acceptable usage for generative AI.

Areas that you need to consider covering off include the use of content generated by AI.  Think of it this way; many clients assume that they are communicating with or receiving information from a human, particularly in the area of consulting or content creation.  While you may wish to use AI to build a framework for client work, there is still a need to ensure that it doesn’t replace experience and expert knowledge that is applied to the individual circumstances of your clients. And really importantly, it should not be used to replace the professional skills and experience of employees and other appropriate research tools for undertaking research and fact checking activities.

Hallucinations and misinformation

A little about the use of information that AI generates……. Generative AI has the ability to hallucinate and while that sounds like someone at the backend is dropping LSD and punching out fun facts, this is where the information generated sounds spot on but the reference sources and the actual detail are completely made up.  The AI (or Large Language Model) just does what it is told to do – it doesn’t have a concept of right and wrong. It just creates the content you asked for.

Take for example this little hallucination that ChapGPT came up with when asked “what is the world record for crossing the English channel entirely on foot”, it replied: “The world record for crossing the English Channel entirely on foot is held by Christof Wandratsch of Germany, who completed the crossing in 14 hours and 51 minutes on August 14, 2020.”  I think we all know that the English Channel is a fairly large body of water so crossing it by foot is ridiculous.  Also, Mr Wandratsch actually swam it in 7 hours and 3 minutes in 2005 which was a record at the time.  Talk about fake news!

False information about something so obviously silly is not important in real world terms but how about the lawyers who presented AI hallucinated legal research before the court.  The lawyers, naively, believed they had found a number of cases that supported their case thinking that AI was able to access sources not available to the lawyers.  Suffice to say, they have an angry judge and they look unprofessional because they didn’t check sources and complete thorough research.  Oops.    

Minimising risk

There are also implications around intellectual property, confidentiality and security.  You must be careful about what you enter into any online AI engine so as to avoid breaching any of these areas and exposing confidential information. Any data you enter into an AI system is shared with the company that operates that generator and can form part of that generator’s training data.  This creates the risk that the data may be exposed to other users of the AI program and create significant issues for your business.

Industries such as health and finance have extremely strict regulations regarding the handling of customer data, and when using AI systems all businesses should apply similar stringent protocols. Outlining what your employees may or may not enter into an AI system is very important, as is directing your employees on what settings they should apply on their computers and where, when and why they can use AI in their roles.

Without doubt, AI is going to have a major influence on the way we work in the future with companies like Google, Microsoft and Apple investing heavily in this space guaranteeing it will move faster than governments can legislate.  Getting out on the front foot to ensure your business is using this technology appropriately will help mitigate risk as the technology develops.

If you are unsure as to where to start, HR Staff n’ Stuff can help you establish an AI policy to suit your business needs.