Have you been keeping up with new AI tools like MidJourney and ChatGPT?
This new processing technology is equal parts terrifying and exciting. Essentially, they are tools capable of creating new content like images and text. Instead of needing people to make art and write things, you can ask an algorithm to do it. Yikes.
ChatGPT is the trendiest tool on the block right now. This new language processor from OpenAI can generate human-like text responses for use in chatbots and articles. With the right prompts, ChatGPT will automatically generate accurate, well-written content—that's arguably of a higher quality than many of the human-authored clickbait articles littered across Google and social media.
Over time, you train the AI to answer questions and write things for you. It's an awesome tool, but experts are already creating screeds of guidance for users around one key issue: prompts.
Here's how it works: Solving a problem has several distinct stages, and ChatGPT is an excellent tool for some of them.
5 stages of solving a problem
Here are the five basic stages of solving a problem:
1. Identify the problem
2. Articulate the problem
3. Generate potential solutions to the problem
4. Evaluate each solution for its potential effectiveness.
5. Implement the most effective solution you have selected.
How ChatGPT helps to solve problems
The first two components of this process are identification and description. All problem-solving requires us first to understand what's going wrong and then articulate and frame it in a useful way. Then, we can generate solutions. ChatGPT is not much help in identifying and explaining what's going wrong in the first place.
Then, we need to generate potential solutions, drawing on precedent, research, and innovation. ChatGPT is a great tool to support this step and can help to create a more comprehensive list than you might have compiled alone - though it can struggle to be creative or identify solutions from other sectors and topics, unless you specifically ask it to do so.
Next, those solutions are evaluated for quality and relevance, considering our circumstances, constraints and capabilities. ChatGPT can help you with this bit, too, if you're clear about the parameters of quality.
Finally, and most importantly, your solutions need to be implemented, and regularly adjusted in response to changes and concerns. ChatGPT is awesome, but unfortunately, it cannot do the actual work for you - that article still needs checking, tweaking and posting.
How to make the most of ChatGPT
The main limitation of a language AI is the person using it and their ability to think clearly, evaluate carefully and act decisively. The first step to this is your questions: if you can't ask good questions, the tool can't give you good answers.
For ChatGPT, the guidance is clear:
- Have an idea of what your purpose is and the kind of result you're after
- Develop precise, clear, short sentences that point toward that response
- Explain the context of the question inside the question
- Pay careful attention to your word choice so ChatGPT can correctly assess what you mean
- Avoid asking questions with yes/no answers
- Avoid very general questions.
This is useful advice for asking questions in any context, AI or not.
No matter how clever technology gets, it still needs one important input to work well: clear thinking expressed as quality questions. Regardless of your job title or industry, the same applies to you.
Quality questions are the most important skill
Over time, many of the things you currently add value to your employer or clients will be automated. Whether you're an accountant, a salesperson, a writer or a coder, there will soon be an algorithm that can solve many of your problems - and they'll probably do it faster, more accurately and with less faffing around.
But the parts of your job that AI can't replicate involve strategic, creative, and critical thinking skills and your ability to execute and change course. To get the most out of a tool like ChatGPT, you must master connecting disparate ideas, seeing new opportunities, understanding how things fit together, experimenting and, most importantly, asking good questions.
If you're not currently confident in your ability to ask good questions—whether of your boss, your team, in meetings, or using a chatbot prompt—it's time to learn.
Stay tuned for the next few weeks as I write a series on how you can ask better questions so you can get better answers.
Til next week,
A
Further reading
Alicia McKay
