Limitations of AI

Artificial intelligence (AI) can do things that not long ago would have seemed like miracles or science fiction. Dr Amanda Williamson, Senior Lecturer in Innovation and Strategy at the University of Waikato and a Manager in AI & Data Consultancy at Deloitte, cautions us to look closer so the limits of the tools – functions of how they were created and by whom – become apparent. We need to be aware of these limitations if we don’t want to see AI tools misused.

Questions for discussion

• Amanda gives an example of biased data. Can you think of an example of how data could be biased?

• How do you think we could help assure generative AI applications are not biased?

Transcript

Dr Amanda Williamson

If we’re not taught about the limitations of these tools then we’re going to run into real trouble, because there are three things that these generative AI tools do.

The first one is they hallucinate. That’s a fancy way of saying ‘confidently b*******s’. You know, it’s 100% confident but not always correct.

The second one is it’s built on toxic and biased internet data. So when you build it on that garbage in, garbage out, you’re going to get biased information and toxic things that you might not even realise is coming out there. So if you ask Bing Chat, for example, to create an image of a computer scientist for you, it’s very likely to create a middle-aged white man. You know, that’s just what it does.

And then the third issue is copyright and IP issues. You might be using it to create a blog post or an essay or even create a report to use in a corporate setting. And that information is not yours. Often it’s going to be someone else’s words, someone else’s ideas.

Acknowledgements

Dr Amanda Williamson, Senior Lecturer in Innovation and Strategy at the University of Waikato and Manager in AI & Data Consultancy at Deloitte.

Video footage courtesy of the University of Waikato. The video is from a recording of the Kaupapa Kōrero panel discussion Bots vs Beings: How Will AI Impact Your Life and Work? held at the University in June 2023.

The full discussion with Professor Mike Duke, Dr Amanda Williamson, Professor Nick Agar and Associate Professor Te Taka Keegan and facilitated by Professor Bryony James can be viewed here.