Chat GPT (Part 1) – Experimentally

Have you tried it? Were you impressed?

Using artificial intelligence to find things out is undeniably fun. You can sign up to Chat GPT’s free service to ‘play’ with it, and ask it all sorts of questions. But although you might think it’s just an advancement on Google, enabling you to type a question in a box and get what looks like an intelligent written reply, there are pitfalls as well as advantages. I’ve done a little experimentation this week, on topics that interest me.

Scottish Song Collecting?

When I first heard about it, at Christmas last year, I immediately asked it a question about something I knew a lot about, Scottish song-collecting. It named a nineteenth-century society that I’d never heard of. ‘But Mum!’, laughed my family, ‘you wrote the book about it. And you haven’t even heard of that society?’

Ashamed, I spent a few hours trying to find the confounded society. And you know what? It doesn’t exist. I’ve seen it said that Chat GPT is simply set up to provide what looks like a reasonable answer to any question that it’s asked. The ‘society’ that it named would have been a reasonable name for a society, if the terminology was being used in the nineteenth century. But it wasn’t!

What’s new?

I know, too, that the free version is based on the internet at a particular point in time, so if you’re trying to find out something very current, it won’t work. I tried again yesterday, asking about something I thought colleagues might be discussing. But of course, it doesn’t know what’s being discussed in 2023, so I got no reply.

So today, I had another go.

Explain why libraries are acquiring more music by women

I was very impressed this time. It even listed five key factors: ‘Promoting gender equality … Broadening Cultural Representation … Recognising underrepresented talent … Meeting user demand and interests … Educational and research purposes …’

If I’d written that answer, it would certainly have passed muster in a blog post. So, if Chat GPT looks at what people have said in response to that question, and creates an answer in line with those observations, then the job’s done and dusted. However, if it had been presented as an answer to an essay question, an essay-marker could have been fooled into thinking it had been carefully researched and considered. Or could they? There were no references, no named authorities or published sources. Not so good, then. (Apart from which, if I had set an essay that was going to be assessed, I’d like to think the author had done a bit of research of their own, and cited a few opinions by respected experts in the fields.)

I asked Chat GPT where they got the info from.

Where did you get this information?

Chat GPT is honest!

As an AI language model, my responses are generated based on a mixture of licensed data, data created by human trainers, and publicly available data. I have not been directly trained on specific libraries or their acquisition policies. My previous response was based on my understanding of the topic as a language model trained on a diverse range of text, including books, articles, websites, and other written sources.

When providing information, I strive to offer accurate and up-to-date content. However, it’s important to note that my responses may not always reflect the most current news or events. The field of library acquisitions is constantly evolving, and it’s always a good idea to consult the latest research, publications, or speak with professionals in the library field for the most accurate and specific information.

Chat.openAI.com

Well, that makes sense. But I thought it was worth pushing just a little further.

Can you cite any sources for this data?

I apologize for any confusion, but as an AI language model, I don’t have direct access to a database of sources or the ability to browse the internet. I don’t have citations to provide for the specific data mentioned in my previous response.

I can tell you that the information provided is based on general knowledge and understanding of the topic as of my last training in September 2021. The points mentioned are commonly discussed aspects related to diversity and representation in library collections.

If you’re looking for specific sources or more in-depth information on this topic, I would recommend consulting scholarly articles, books, or reports on library acquisitions, diversity in music collections, and gender representation in the arts. This will help you find more detailed and sourced information to support your research.

Chat.openAI.com

So, I would see it as a bit like Wikipedia – it’s great for giving an overview of a subject, but it doesn’t take away the need to do your own homework! It makes things up (the imaginary 19th century society), and it doesn’t cite sources. Moreover, if you use the free version, you won’t get the most up-to-date data or opinions.

And I’d also be concerned to know how a teacher would be able to tell that an open AI source had been used. Someone could copy and paste the answer, submit it, and never have actually done any deep thinking, or achieved any level of learning.

It’s something I’d like to know more about! There’s no point in advising, ‘don’t use Chat GPT’. It’s there, and it won’t go away. But we should certainly be advising students to use it responsibly. It might be tempting to use it, but not as a replacement for doing the work!

What about Writing Research Papers? Helpful or Unethical?

On a related topic, have you heard about the most prolific scientist in the world (allegedly), producing articles on average every 37 hours in the present year 2023? With the assistance of AI, I imagine, but it can’t be the Chat GPT that we ‘ordinary’ folk are playing with. After all, scientists have to be slap-bang up to date,for one thing. And also, I presume he is feeding data INTO his AI, otherwise how will he get new results and new discoveries? Intriguing. Let me say that I’m not entirely clear whether the suspension is at all connected with the use of AI – it seems more to be about who paid him (or didn’t, depending who you believe) for some of his work, and that he has been punished for something by his university. The fact that he is so prolific seems to be a side-issue here, even though that’s the aspect that caught my attention.

Here’s the citation for that particular story:-

‘One of the world’s most cited scientists, Rafael Luque, suspended without pay for 13 years.’

Ansede, Manuel, for EL PAÍS English. 2 April 2023

Truly, these are interesting times. Future scholars will look at my generation and marvel at how comparatively innocent we were, and how slow to get results!