sourcegraph
March 28, 2024

Welcome back to On Tech: AI, a pop-up newsletter introducing you to 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 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 learn how to realize AI’s full potential.

People from all walks of life—students, programmers, artists, and accountants—are experimenting with how to use artificial intelligence tools. Employers are posting job listings, looking for people who are good at using them. Soon, if you haven’t already, you’ll have the opportunity to use AI to simplify and improve your work and personal life.

As The Times’ personal tech columnist, I’m here to help you understand how to use these tools safely and responsibly to improve many aspects of your life.

In today’s newsletter I’ll be discussing two general approaches that are useful in many situations.

Then, over the next few weeks, I’ll offer more specific advice for different areas of your life, including parenting and home life, work, organization of your personal life, study/education, creativity, and shopping.

A few common sense caveats to start with:

  • If you are concerned about privacy, please omit personal details such as your name and place of work. Tech companies say your data is used to train their systems, which means others could see your messages.

  • Do not share confidential data. Your employer may have specific guidelines or restrictions, but in general it is a very bad idea to enter trade secrets or sensitive information.

  • Illusion: Chatbots are powered by a technique known as Large-Scale Language Models, or LLMs, which derive their abilities from analyzing large amounts of digital text culled from the internet. A lot of things on the web are false, and chatbots may repeat these lies. Sometimes, they can make things up when trying to predict patterns from their vast training data.

Chat GPT, must and poet is one of the most popular AI chatbots. (To use ChatGPT, you need to create an OpenAI account and subscribe to its most advanced version. Bing requires you to use Microsoft’s Edge web browser. Bard requires a Google account.)

Even though they look simple and easy – you type in a box and get an answer! — Asking questions in the wrong way can produce general, unhelpful, and sometimes downright wrong answers.

It turns out there is an art to typing in the exact words and composition to generate the most useful answers. I call these golden tips.

People who are getting the most out of chatbots have been using variations of these tactics:

“Act as if.” Starting your prompt with these magic words will instruct the robot to imitate the experts. For example, typing “as if you were an SAT tutor” or “as if you were a personal trainer” would direct the bot to model people around those professions.

These hints provide additional context for the AI ​​to generate responses. AI doesn’t actually understand what it means to be a mentor or a personal trainer. Instead, hints are helping the AI ​​exploit specific statistical patterns in its training data.

Weak cues without guidance will yield less useful results. If all you type in is “what should I eat this week?” the chatbot will come up with a generic list of meals for a balanced diet, such as a turkey stir fry with colorful veggies for dinner (which sounds “meh” to me. ”).

“Tell me what else you need to do.” For more personalized results — for example, health advice tailored to your specific body type or medical condition — ask the bot to request more information.

In the personal trainer example, the prompt could be: “As if you were my personal trainer. Create a weekly workout and meal plan for me. Tell me what else you need to do to make this happen.” Then, the robot You may be asked about your age, height, weight, dietary restrictions, and health goals to create a custom week-long meal plan and fitness plan for you.

If you don’t get good answers on your first try, don’t give up right away. Even better, in the words of Ethan Mollick, a professor at the Wharton School of the University of Pennsylvania, Treat Robots Like Human Interns: “When it makes a mistake, point it out and ask it to do better.” Be tolerant and patient, and you may get better results.

After you master the tips, you can make your chatbot more helpful over time. The key here is to avoid thinking of your chatbot as a web search and start with a new query each time. Instead, keep several conversation threads open and add to them over time.

This strategy is the simplest for ChatGPT. Bing requires you to periodically reset your conversations, while Bard doesn’t make it as easy to jump between conversation threads.

Sydney, Australia-based entrepreneur Natalie Choprasert, who advises companies on how to use AI, uses ChatGPT as a business coach and administrative assistant. For each of these characters, she keeps separate dialogue going side-by-side.

For the topic of business coaching, she shared insights about her professional background as well as company goals and issues. For the executive assistant thread, she shares scheduling information, such as clients she meets with.

“It builds and trains correctly, so when I ask it a question later, it comes up in the right context and gives me close to the answer I’m looking for,” Choprasert said.

She shares an additional golden tip that has trained her assistants to be more helpful: apply frames. She recently read A Clockwork, a book about entrepreneurship. When she asked ChatGPT-the-business-coach for advice using the “Clockwork” framework, she was pleased to see that it could incorporate the book’s principles into her plans of action for expanding her company.


What are some golden tips that got you the most impressive and helpful results from artificial intelligence? Email us your example. We may use your submissions in future editions of this newsletter.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *