ChatGPT: The Art of AI-ssistance?

Photo by Alexander Sinn on Unsplash

I’m no expert on AI, but I am fascinated by it, and I’ve spend many hours of my free time (when I could be doing something more useful like reading a book) playing with the various offerings out there (ChatGPT, Yellow.ai, Deepai, Midjourney, etc), to test out what they can do. And it is impressive, compared with what was possible even just a few years ago.

My brief professional experience of working with AI assistants only taught me just how complex this space is, and how hard it is to do well. The basics of speech recognition (understanding the words you are saying) and Neuro-linguistic understanding (understanding the intent behind those words) have both evolved hugely since then. Things that used to be major problems to solve, such as understanding accents, figuring out who is actually speaking and what the person actually means, are now solved and the space has evolved into thinking beyond mundane tasks (turn my light on, change the channel or order more cheerios) to things that we’ve all maintained only a human can do well.

So, how’s it going?

I’ve chosen to focus on ChatGPT, as it’s become a daily conversation with my team at work. We’re regularly coming up with use cases for where it could be applied, to see how good a job it’s doing. The general conclusion is that it seems to to a half-decent job on a number of things.

e.g. I asked it to come up with a title for this article — something I always struggle with. You can be the judge of whether it did a good job or not.

So, yes — on the face of it, there are lots of things you could get it to do, such as writing SEO optimised content for websites, generating content in multiple languages, handling basic customer interactions…. or suggesting things that it can be used for.

> give some examples of applications for ChatGPT in the travel industry

1. Personalized travel itinerary generation: ChatGPT can be trained on a
database of travel information and destinations to generate personalized
travel itineraries for users based on their preferences, budget, and time
constraints.
2. Virtual travel assistants: ChatGPT can be integrated into a chatbot or
virtual assistant platform to provide users with information on destinations,
transportation options, hotel and flight bookings, and other travel-related
queries.
3. Language translation for travelers: ChatGPT can be used to translate text
in real-time, providing travelers with instant translation services when
traveling in a foreign country.
4. Travel content generation: ChatGPT can be trained on a database of
travel-related content to generate articles, blog posts, and other types of
travel content on various destinations and travel-related topics.
5. Social media management: ChatGPT can be used to generate responses to
customer inquiries and complaints on social media platforms, allowing travel
companies to improve their customer service and engagement.
6. Travel chatbot: ChatGPT can be integrated into a chatbot that can help
travelers plan their trip, answer their queries and help them in booking
their tickets and accommodations.
7. Personalized travel recommendations: ChatGPT can be trained on data of
user's preferences and travel history to provide tailored recommendations for
destinations, activities, and accommodations based on their individual
preferences.

The problem is, everyone else is doing the same thing. Whilst it might be harder to spot AI generated content from human content, to the uninitiated, people will soon learn how to spot the difference and will react accordingly. Content will start to become more vanilla and lack diversity, and as a consequence be less engaging, feel less human.

And while humans might struggle to spot the difference initially, machines do not. Universities and schools are using AI to spot AI generated essays and coursework (the irony!). Google have already stated that AI content is against their guidelines, so you can expect that human generated content will perform better in search rankings. Lots of forums have banned AI generated content. Why have they done that? Because they are trying to prevent quality being replaced by quantity.

There is one thing though that’s bothering me, beyond all of this. The content being generated comes from all of us, from years of scouring the internet for the knowledge (the big data) that is being used as the basis for this. That’s the reason why ChatGPTs knowledge is currently limited to 2021 data (try asking it about anything that happened in 2022!).

So what happens when that content gets gradually diluted by AI generated content. In theory, AI will ignore AI generated content, but it does suggest that the quality and diversity of content will gradually decline.

The art generated by Stability AI, Midjourney and Deviant Art is based on analysis of “real” art generated by real artists, hence why some of these companies are being hit by copyrighting class action law suits by groups of artists. Without the “real” art to base any AI generated content on, would the quality be the same?

Midjourney: Imagine “a gang of hells angel pigs on motorbikes”

So, do I think there are a multitude of really interesting use cases for ChatGPT?

Yes, I do. And the likes of Microsoft clearly think the same, with their recent announcement of a multi-billion dollar investment in this space.

Do I think it will do us all out of a job any time soon?

No, but clearly AI is improving it’s ability to be more human, that extends the range of things it is capable of doing well.

However, if you want to test how well AI can replicate the quirkiness of what it is to be human, ask it to generate a joke. Yes, exactly. There’s a reason why Amazon hired comedians to write the jokes that Alexa comes out with.