Since November 2022, there’s been a new internet sensation in town: ChatGPT. As a generative AI tool, Chat GPT can create content from news-type articles to brand-new jokes — although whether they’re funny or not is, of course, entirely subjective. AI-powered content creation and conversational skills have a number of uses for businesses, such as automated customer service. But how will your kids engage with these new, more human-like types of chatbots — and should you be worried? Or will the benefits outweigh the risks?
What is ChatGPT?
Chatbots are AI-powered tools designed to answer questions or respond to queries in pre-programmed. For example, you might try and buy some furniture and get automated responses about bed dimensions or shop opening times. ChatGPT has expanded upon that concept massively, with the capability to answer just about any question you can think of.
You can read an example of a “news” article written by ChatGPT here.
The AI uses statistical probability to figure out what words can and should go before and after other words to form meaningful responses to queries. Think about the texting options on your smartphone: when you click an auto-generated word, another word will appear, then another. This AI can do the same but far more accurately and on a far larger scale.
What is OpenAI?
ChatGPT’s creator, OpenAI, is a California-based research lab. They created the predecessor to ChatGPT, called GPT-3. Both versions use an AI concept called a Large Language Model or LLM. This means the AI is provided with, or “fed”, billions of examples of writing and conversations, including online articles and thousands of books.
OpenAI have been transparent with their AI training methods, publicly documenting queries they gave the AI, how it responded, and how they fine-tuned it. For example, they asked the AI about a piece of code, and showed how the AI asked for more context and clarification, as well as providing a potential solution.
Potential benefits of ChatGPT
A fluent, human-like content generating AI could potentially provide the following benefits:
· Make article and blog generation faster for individuals and organizations
· Create entirely personalized responses to individuals within a large group, such as a class of children
· It can be used to create lessons plans, meeting agendas, and other basic lists — these should still be sense checked by a human though!
· Create talking points or questions to prompt debate on a range of subjects
· Replace automated responses to customer service or tech support queries with real-time answers tailored to the specifics of each question
The Risks
Because AI uses statistics to create its sentences, it doesn’t always get things right. In fact, sometimes it’s wildly off the mark. For example, ChatGPT told a user that the only country in the world that starts and ends with the same letter is Chad. To a human, that’s obviously wrong. But the AI doesn’t recognise that.
For kids (and adults), this leaves them susceptible to misinformation and potentially unsure what’s fact and what’s fiction.
A common concern about AI is that it will take jobs or even make areas like customer service redundant. However, as AI is still fallible and prone to errors, this is unlikely. Currently, it’s a tool to aid those skills, rather than a human replacement.
The most dangerous risk is one that’s been present in all form of AI so far: bias. AI is only as neutral as its creators. The more it learns, the more it leans towards the existing biases of AI trainers. This can lead to unfortunate racist, sexist, or other discriminatory statements or content. OpenAI states they’ve tried hard to keep the AI away from bigotry, but users have still found ways to “make” the AI say inflammatory things.
Warning our kids about the danger
Kids should understand that, as of right now, they can’t rely on ChatGPT or other AI or machine learning tools to do their homework for them. There’s no guarantee the responses will be correct or even sensical.
It’s important to warn kids that all AI tools are ultimately susceptible to human input, and could say upsetting or even bigoted things. Teaching kids tolerance and discernment provides them with tools to navigate the rapidly expanding world of AI. Getting them involved in tech education groups or events from a young age could help them understand AI at a more fundamental level. This makes them less likely to be duped by inaccuracies, misinformation, or discriminatory AI-created content.
Currently, ChatGPT is so popular that if you try to access the tool to ask it your own questions, you might get a “ChatGPT is at capacity” notice. You might also be treated to some of its own AI-generated jokes: “The ChatGPT servers must be stuck in traffic… in the internet highway!” What do you think? As funny as a human being? Or no difference? Regardless of how fluent AI chatbots get, they can’t ever replace the compassion and warmth of genuine human interactions, but they could, in time and with careful ethical guidance, become genuinely useful tools for education and communication.
Sign up here to our Girls Into Coding mailing list to be among the first to receive our articles and to know about future events!
