Welcome back to On Tech: AI, a pop-up newsletter dedicated to informing you about artificial intelligence, how it works, and how to use it.
A few months ago, my colleagues Cade Metz and Kevin Roose explained the inner workings of the AI., including chatbots like OpenAI’s ChatGPT, Microsoft’s Bing, and Google’s Bard. Now we’re back with a new mission: to help you unleash the full potential of AI.
People from all walks of life – students, programmers, artists and accountants – are experimenting with the use of AI tools. Employers post vacancies and look for people who know how to use them. Soon, if not already, you will have the opportunity to use AI to streamline and improve your work and personal life.
As the Times personal tech columnist, I’m here to help you figure out how to safely and responsibly use these tools to improve many areas of your life.
I’m going to spend today’s newsletter talking about two general approaches that will be useful in a number of situations.
Then, in the coming weeks, I will give you more specific advice on different aspects of your life, including parenting and family life, work, organizing your personal life, learning/education, creativity, and shopping.
A few sensible warnings to start with:
-
If you have privacy concerns, omit personal information like your name and work location. The technology companies state that your data is used to train their systems, which means that other people could potentially see your information.
-
Do not share confidential information. Your employer may have specific policies or restrictions, but in general, entering trade secrets or sensitive information is a very bad idea.
-
Hallucinations: Chatbots are based on a technology called a large language model, or LLM, which derives its skills from analyzing enormous amounts of digital text from the Internet. A lot of things on the internet are false and chatbots can repeat these untruths. Sometimes while trying to predict patterns from their extensive training data, they can invent things.
The Golden Prompts
ChatGPT, Bing And bard are among the most popular AI chatbots. (To use ChatGPT you must create an OpenAI account and the most advanced version requires a subscription. Bing requires the use of Microsoft’s Edge web browser. Bard requires a Google account.)
Although they look easy to use – you type in a field and get answers! – Asking questions wrong leads to general, unhelpful and sometimes completely wrong answers.
It turns out there is an art to typing in the precise words and phrasing them in a way that generates the most helpful answers. I call this the golden prompts.
The people who make the most of chatbots use variants of these strategies:
“Pretend.” Starting your prompt with those magic words instructs the bot to emulate an expert. For example, typing “act like you’re a tutor for the SATs” or “act like you’re a personal trainer” will direct the bots to orient themselves towards people in those professions.
These prompts provide additional context for the AI to generate its response. The AI doesn’t really understand what it means to be a tutor or a personal trainer. Instead, the prompt helps the AI refer to specific statistical patterns in its training data.
A weak prompt with no guidance will produce less than helpful results. If you just type “What should I eat this week?”? The chatbot creates a general list of meals for a balanced diet, such as stir-fried turkey with a side of colorful vegetables for dinner (which sounds very “meh” to me).
“Tell me what else you need to do this.” To get more personalized results — like health advice for your specific body type or health condition — invite the bot to request more information.
In the example of a personal trainer, a prompt might be, “Act like you’re my personal trainer.” Create a weekly exercise program and meal plan for me. Tell me what else you need to do to do that.” The bot may then ask for your age, height, weight, dietary restrictions, and health goals to create a week-long meal and fitness plan for you.
If you don’t get good answers on the first try, don’t give up. Better yet, in the words of Ethan Mollick, a professor at the University of Pennsylvania’s Wharton School: Treat the bot like a human intern: “If it makes a mistake, point it out and demand that it get better.” Be forgiving and patient, and you’ll probably get better results.
Link your chatbot conversations
After you understand the prompts, you can make your chatbot more helpful over time. The key here is to not treat your chatbot like a web search and start with a new request each time. Instead, keep multiple conversation threads open and add to them over time.
This strategy is easiest with ChatGPT. With Bing, you need to reset your conversations regularly, and Bard doesn’t make it that easy to switch between conversation threads.
Natalie Choprasert, an entrepreneur from Sydney, Australia who advises companies on the use of AI, uses ChatGPT as a business coach and executive assistant. She conducts separate side-by-side conversations for each of these roles.
In the business coach thread, she provides insights into her professional background and the company’s goals and problems. For the Executive Assistant thread, she shares appointment information, such as the clients she is meeting with.
“It builds up right and trains right. So when I ask him a question later, it’s in the right context and it gives me answers that are close to what I want,” Choprasert said.
She shared a golden bonus prompt that has trained her assistants to be extra helpful: Apply a framework. She recently read Clockwork, a book about starting a business. When she approached ChatGPT-the-business-coach for a Clockwork consultation, she was pleased to see that it enabled principles from the book to be incorporated into an action plan for expanding her business.
Share your suggestions
What are your golden pulses that have given you the most impressive and helpful results through AI? Email us your examples. We may use your submissions in future editions of this newsletter.
