Since ChatGPT launched last year, millions of people have used AI for everything from learning languages to writing silly poetry.
With an emerging technology, there’s a lot of hype. It can be hard to tell what to use it for.
Fortunately, there are many ways you can use AI to help you learn and acquire knowledge efficiently.
AI and education can work together. But it’s important to understand a little bit about how AI works, to understand its limitations.
Here’s what we’ll cover:
- How Generative AI Works
- Using AI for Research
- Using AI for Language Learning
- Using AI for Brainstorming
How Generative AI Works
Before we get into strategies for using generative AI, it’s a good idea to get a good understanding of how the technology actually works. This will give you a better sense of what it really can and cannot do.
Unfortunately, there is a lot of hype and misinformation about AI. Part of that is just a side effect of marketing.
Many people are under the false impression that there’s something like “a little person” inside the computer that’s doing the writing. That’s not how it works.
Basically, text generating AI programs work like the autocomplete on your phone: by guessing the next word. It starts with what you give it, and keeps guessing a plausible next word until a few hundred words are written.
At no point in this process does the program do anything to fact check what it writes.
But of course, AI can do much more than ordinary autocomplete. It can make grocery lists, and re-write Hamlet with a pirate accent!
But how does it do that?
Again, there’s nobody inside the machine actually thinking about this. Instead, the computer relies on billions of statistics compiled about words.
For example, suppose you had the words “the United.” The next word would likely be “States” or “Kingdom,” but probably not “Potato.”
AI firms collect tons and tons of data to train their language models. ChatGPT was trained on over 570 gigabytes of text information.
It has probably seen “the United States” and “the United Kingdom” millions of times, but never “the United Potato.” So it knows that “the United Potato” is an unlikely thing to write.
If a generative AI program sees the words “the United,” it can guess the next word based on the previous words.
For example, if the paragraph before included words like “biscuit” and “queen,” then “Kingdom” would be a good guess. If the paragraph included “hamburger” and “president,” then “States” would be a better guess.
When you ask ChatGPT to write Hamlet with a pirate accent, it already has statistics compiled about pirate words and about Hamlet. So it simply adjusts its output to reflect that.
Of course, the exact details of the program are more complicated. But this should give you a rough idea of what’s actually going on underneath the hood.
Using AI For Research
AI is a double edged sword when it comes to research. It is an incredible tool for finding general information on a large number of topics. But it also just… makes stuff up sometimes.
Again, there’s nothing inside the program that’s doing any fact checking.
Because of this, for most subjects, generative AI is probably not any better than Wikipedia. At least Wikipedia has people trying to keep it honest.
In fact, ChatGPT was trained on thousands of Wikipedia articles, which explains why it sounds like Wikipedia much of the time.
For research purposes, keep the following in mind:
- AI is not a source.
- You need to check everything AI says.
- Use AI as a research starting point, not an end.
- AI can make up sources that don’t exist.
AI is not a source
From an academic standpoint, AI does not count as a source. You cannot cite it in a research paper.
This is because there’s no one inside the program doing any fact checking. For an academic book or article, someone actually went out and studied things to produce it.
You need to check everything AI says
This is the unfortunate truth about AI. While most of the time it probably will make accurate claims, there is simply no way to tell the difference without checking yourself.
Use AI as a research starting point, not an end point
AI is a great way to find out what ideas are nearby what you want to learn about.
For example, suppose you ask “What was World War II?” The output will probably contain references to events like Pearl Harbor and D-Day, which you can then research on your own.
AI can make up sources that don’t exist
This is one of the most deceptive aspects of current AI tools. If you ask ChatGPT to cite its sources, it can make up links that do not exist.
This is particularly problematic if you looked at the links and simply assumed they were real. Always double check any links it provides.
ChatGPT is not currently connected to the internet in such a way that it can find links to justify its claims, but perhaps that could change in the future.
Using AI as a language learning tool
AI has more potential value as an educational tool. This is because when learning a language, you often simulate interactions that don’t have to be real in the first place. So it doesn’t matter if the AI makes stuff up- it’s supposed to.
For example, you can start a conversation with “Pretend you are a fast food worker. Take my order.” And then have a conversation with it.
There are endless possibilities here:
- Pretend you are my boss at work. Give me an assignment.
- Pretend you are my English teacher. Give me a lesson.
- Imagine you are a travel guide. Recommend the best hidden gems in a specific city.
- You are a wise philosopher. Share your insights on life and its meaning.
- Let’s pretend you are a sports commentator. Provide live commentary for an imaginary game.
- Pretend you are my math teacher. Give me a problem.
- You are a nutritionist. Advise me on healthy eating habits and create a personalized meal plan.
- Pretend you are Iron Man. I will interview you about your latest invention.
- Imagine you are a fashion stylist. Help me put together a stylish outfit for a special occasion.
- Pretend you are a career counselor. Give me advice on finding the perfect job.
- You are a scientist on a mission to Mars. Describe the challenges and discoveries of space exploration.
- Pretend you are a detective. Help me solve a mysterious case.
- Imagine you are a therapist. Provide guidance and support for overcoming challenges.
- Pretend you are a chef. Share your favorite recipes and cooking tips.
You can make these prompts as detailed and specific as you like. You can play around with your expanding English knowledge to add personality characteristics, like funny, quiet, or analytical.
So long as you stick to non-existent people, there is no risk of false information. However, chatbots can contradict themselves, and they can say things that go against canon in a fictional universe.
Although AI chatbots can be great language learning tools, they are not substitutes for real human teachers. That’s right, you should still get that TEFL if you’re thinking of teaching overseas!
Using AI for Brainstorming
Brainstorming is another activity which relies more on imagination than on facts. That makes it perfect for AI.
Most people find that the best way to come up with ideas is to think for a little while, then give up and go for a walk or take a shower. Miraculously, an idea comes to you when you aren’t trying to think of one.
What really happens is your subconscious mind just takes everything you were thinking about and mashes it together into something new while you were distracted.
In a weird way, AI language models work a little bit like this. Since they are based on probability, they effectively just mash ideas together.
You can use AI to speed up the time it takes to incubate new ideas with prompts like these:
- Write some suggestions for new business ideas.
- Come up with some fun activities to do at a party.
- Give me some ways to budget money better.
- What are some good ways to prompt ChatGPT?
Conclusion: The Future is Open Ended
Now is an exciting time for AI. But it remains to be seen exactly how useful it will become.
Most new technologies go through a phase of inflated expectations, when people try to use them for many different things. As time goes by, we figure out exactly what they are really good at.
Generative AI is highly open ended. It’s up to you to figure out how you want to use it, and what you want to use it for.
Don’t be afraid to experiment, and to share the results you get. But do be critical about what AI chatbots say, and fact check what you read for accuracy.