Welcome To Day 2 Of Our AI Life Hacks Series |
Now that we have the introductions out of the way, think of this email as your first day of AI class. Don’t worry about having a fresh batch of sharpened pencils—all of our work is going to be done on the computer (or your phone or tablet, if that’s what you prefer).
Figuring out how to communicate with large language model chatbots is like learning a new language: It’s not going to make sense all at once. Sure, you can enter any prompt into a chatbot and it will respond with something, but it takes practice and time to learn how to properly word your inputs for a variety of situations in order to maximize the results. In fact, Forbes contributor Aytekin Tank even asked ChatGPT for advice on how to learn to write prompts. Here’s how the chatbot responded:
“The most effective way to learn is through practice. Regularly using and tweaking prompts will help you understand what works best for different contexts.”
Over the course of this series, we will offer text that you can copy and paste directly into your chatbot of choice, but remember that the work doesn’t end there. The only way to fully understand the capabilities of this booming technology is to keep digging and make adjustments. Use the ideas in this course as a starting point and then experiment and brainstorm new ideas to test the power of AI.
Before we dive into any actual prompts, let’s do a quick crash course on how this technology grew so quickly, and the major players in the game. |
|
| What Are The Biggest Large Language Models? |
|
While we could trace the history of chatbots back to the 1960s, when computer scientist and MIT professor Joseph Weizenbaum developed the Eliza model at the school’s Artificial Intelligence Laboratory, in this newsletter we’ll focus on the more recent rise of large language models like OpenAI’s ChatGPT.
You’ve probably already interacted with a rudimentary chatbot—think of the customer service chats offered by online business storefronts or your insurance website that can sometimes answer the questions you ask. Large language models (LLMs) take things a few steps further. These sophisticated language machines can understand “context, nuances, and complex sentence structures,” writes Forbes contributor Sunil Rajaraman. “They can write different kinds of creative content, translate languages, analyze sentiment and even answer your questions in an informative way.”
LLMs are trained on huge datasets of text and code, which allows them to perform a variety of different tasks. You can use an LLM to summarize a text document or write an email (as we’ll discuss later in this course!), or you can ask it to perform more complex tasks such as writing and editing code like HTML or CSS. And their capabilities are growing all the time.
Here’s a quick breakdown of some of the bigger names in the world of large language models: |
|
 | On November 30, 2022, OpenAI released ChatGPT to the general public. In many ways, the world changed forever that day. OpenAI constantly works to update and improve upon its chatbot, which can create and transform images, generate infographics, give feedback on how to improve a website and more. Following ChatGPT’s release, a number of LLMs quickly followed, each with their own strengths. |
|
| Facebook parent company Meta offers its Llama 4 model, which differs from other models in that it is (at least partially) open source. This means that the code and training materials are made available to the public, which enables some of the more tech-savvy users to create their own specialized chatbots. |
|
| Another major player in the game is Claude, which was created by Anthropic, a company founded by former OpenAI employees. Anthropic “pitches itself as a safety-focused firm that prioritizes the safe and responsible development of artificial intelligence over commercialization and profits.” |
|
| These are just a few of the many advanced chatbots available on the market—and they’re all competing to be faster and more knowledgeable than the others. Most are currently available for free, though they may require that you create an account, and additional features can be accessed for a fee. But this accessibility allows you to take a few for a test drive and decide which feels best. As we make our way through this course, we recommend experimenting with the different LLMs to see how they respond to similar prompts, and take note of what you like and dislike about each. |
|
| A Note On AI Hallucinations |
|
Throughout this course, we will talk extensively about the possibilities of AI and even marvel at its many capabilities. But it’s important to remember that, no matter which LLM you experiment with, AI is not a perfect technology.
One of the biggest problems with AI is its tendency to hallucinate. Put simply, chatbots don’t always tell the truth, and they will sometimes spew out a stream of information that includes “facts” that are seemingly made up out of thin air.
Anyone who works with AI for even a short period of time is bound to come across a few hallucinations—and that’s why it’s important to never take anything an AI creates at face value. Make sure to always double check the outputs you receive—and if the AI cites sources, take the extra step to verify them. This will help ensure that any work you produce with the assistance of AI is mistake-free.
As you experiment with chatbots and begin to incorporate them into your everyday life, don’t forget to be on the lookout for such hallucinations. Blindly trusting an AI to be 100% accurate as it writes a social media post or comes up with an idea for what to eat for dinner can lead to undesired consequences. |
|
| Experiment With A Prompt Or Two |
|
| We’ll be back in your inbox tomorrow with some fun ideas for how you can prompt an AI chatbot, beginning conversations and tackling items on your to-do list. But if you’re chomping at the bit to experiment with some of the chatbots we discussed above, start with something simple, like asking for suggestions. Center your prompt around one of your interests, and begin the conversation. Here’s what we entered into ChatGPT: |
|
| My family is having a movie night this week. We’re interested in the science fiction genre, but are looking for something heartwarming, with little to no violence. Provide 10 possible movie options, with a brief, 1-2 sentence synopsis for each. |
|
The chatbot did exactly what we asked, even noting in a few of the synopses that a film in question contained some adult themes. From there, our options to continue the conversation are pretty open-ended: We could ask the chatbot to narrow down the list, or request more information about a specific movie. Then, we could ask for some fun ideas to make this particular movie night a memorable one … and the possibilities continue from there. We can use the chatbot to come up with trivia questions for a fun pre-movie game or suggestions for costumes that tie to the film we picked.
Whatever you decide to do, have fun and keep pushing the chatbot for more information to see how the conversation evolves. The more detailed you are with your prompt, the better answers you’ll get.
Happy chatting! |
|
|
