I know that the majority of beginners don’t know the basics of AI. Last week I invested 5 hours and $49 to finish Google’s newest beginner AI Essentials course. I made this article to pass along five valuable lessons learned from the course, balance the pros and cons, and give you a straightforward answer on whether or not the certificate you get at the end will actually get you laid—that is, paid—because I need to recoup that cash to fund my parasocial relationship addiction with AI. broke my rule. Since you have a new skill. That was a weird linguistic slip-up.
Takeaway #1: The Three Types of AI Tools
As a whole, there are three types of available AI tools. First, software that AI power uses, that needs very little setup in order to act automatically is what I call a stand-alone utility. General-purpose tools such as ChatGPT, Gemini, Claude, and Perplexity fall into this category alongside niche systems such as Spico, Otter AI, Midjourney, and Gamma. These are all classified as standalone because they can be utilized independently of being integrated into other programs and are available on the websites or apps, with completely different purposes.
The Second Type:
The second type, tools with built-in AI features, is changes that are already within a certain software product. For example, after I’ve drafted a blog post using Google Docs, I can either use the built-in Gemini for Workspace AI tool to edit it or cut and paste the text into a separate program such as ChatGPT to refine it. As a second illustration, I may utilize Gemini for Workspace to generate an image right within Google Slides or use Midjourney as a single, standalone program to produce images for a presentation. Here, Google Docs and Google Slides are integrated AI-enabled products but ChatGPT and Midjourney are single AI solutions.
The Third Type:
Third, a bespoke AI solution is a software application that is specifically designed to address a specific issue. For example, Johns Hopkins University created an AI system with the single purpose of identifying sepsis. This bespoke AI solution enhanced diagnostic accuracy from 2–5% to an average of 40%. And if you’re like me and don’t have a technical background, you might be thinking, “Oh, custom AI solutions are super technical and I’ll likely never have to work with them in the office.” The truth is the opposite—well-constructed custom AI solutions should need minimal to no technical expertise.
For instance, when I was working on the sales team with more than 200 clients to manage quarterly, conducting research on each one of them was clearly time-consuming. Now there are tailored AI solutions that can take in all the data regarding those 200 clients—historical trends, seasonality, industry trends—and prioritize them by how probable it is they would require support so that the salesperson knows how to prioritize their time.
A Quick Warning Before You Buy
By the way, don’t buy Google’s AI Essentials course if you’re considering it. I was only able to discover that you receive the AI Essentials course for free when you sign up for the Google Project Management certification on Coursera, who is graciously funding this segment of the article.
Project Management Plug
Regular viewers are aware that I work full-time, and project management naturally takes up a big portion of my daily life. To be honest, though, I’m largely self-taught, as there wasn’t a default course way back when. But I just began the Project Management certification on Coursera because it’s now essentially the gold standard. Project management is literally universal to all industries and positions, so if you desire to be more efficient in the workplace, click on the link below to sign up for the Google Project Management certification and gain access to the AI Essentials course for free. Thanks, Coursera, for sponsoring this section of the article.
Takeaway #2: Surface the Implied Context
The second thing I learned from the course is a prompt engineering trick: surface the implied context. To demonstrate this, consider your vegetarian friend asking you for restaurant suggestions. You automatically respond with vegetarian options—even though your friend doesn’t say, “Hey idiot, ensure it’s vegetarian.” In this case, the fact that your friend is vegetarian is implied context and must be made explicit when interacting with AI systems such as ChatGPT and Google Gemini.
Here’s an example: you’re preparing to negotiate a raise with your employer. You keep in mind that you received a 10% raise last year, that you’re the company’s best performance this year, and that the industry standard is a 12% raise. You decide to ask for a 15% increase. If you use an AI tool to generate negotiating tactics, you’ll receive a poorer-quality, more generic output if you overlook all that implied information.
If you would like to explore this further, I have a wonderful article on writing the ideal prompt, so I’m going to leave a link to that below. And if you’d like to copy my five productivity prompts, I’m going to leave a link to my entirely free workspace toolkit below.
Takeaway #3: Zero-Shot vs. Few-Shot Prompting
Understand when to utilize zero-shot and few-shot prompting. In short, the term “shot” just refers to examples. Zero-shot implies you are utilizing a prompt with zero examples. One-shot implies that you include an example. Few-shot implies that you include two or more examples.
For example, a zero-shot prompt could be something like, “Write me a pickup line for Bumble”—which is entirely an imaginary situation I would never agree with, much less pursue. A one-shot prompt would be, “Write me a pickup line for Bumble. Use this pickup line my friend employed that served him well,” and you provide an example. A few-shot prompt would appear identical to one-shot, with only two or more examples of successful pickup lines. As should be obvious, the more germane examples you give the AI tool, the more germane the output.
And for the record, if my future wife is watching this, I’ve never actually done this myself. This is just an example for this article. In fact, I don’t even use dating apps.
Takeaway #4: Chain of Thought Prompting
Apply Chain of Thought prompting for difficult tasks. I’ve discussed this topic in earlier articles, but I particularly appreciate this straightforward and simple definition from Google’s course: by breaking one task into smaller steps, you assist the big language model in generating accurate and consistent output.
A familiar real-life example would be composing a cover letter. Option one: you provide your up-to-date résumé and the job advertisement to the chatbot and simply ask it to generate a cover letter for you. Option two: with Chain of Thought prompting, you segment the big task (composing the whole cover letter) into smaller steps. Step one: “On the basis of my résumé and job description, compose an engaging hook.” Then, after making a few adjustments, step two would be pasting that again into the chatbot and having it generate the body paragraph. Rinse and repeat for the final paragraph.
If you’d like evidence that it works, I have a whole article where I’m instructing job hunters on how to use Chain of Thought prompting to create cover letters and enhance their résumés. I’ll provide a link below.
Takeaway #5: Know AI’s Limitations
This is an issue most of us—including me—have a tendency to forget about when employing AI tools: having knowledge of the limitations of AI.
Generally speaking, there are three primary limitations. First, the training data upon which AI models are trained can be biased. If a text-to-image model has only been trained on minimalist graphics. It may not have the capability to create flashy and bold designs. Second, there just wasn’t enough information in the original training data on a particular topic. Most AI models will have a cutoff date. So if you query something that happened recently, it will not have the information to answer correctly. Lastly, hallucinations—factually incorrect AI responses—can happen. Occasionally, this is a desirable feature (for brainstorming). Other times, hallucinations spread misinformation. Always verify AI responses for high-stakes activities—for instance. What type of supplement or vitamin to take based on your health objectives.
Pros and Cons of the Course
Let’s begin with whom this course isn’t for: if you are already leveraging AI tools such as ChatGPT and Google Gemini within your current daily workflow and you’re just trying to delve further into niche AI use cases, this course may not cut it. Even though it clearly elucidates sophisticated matters, examples used are relatively nebulous. One lesson mentioned that a business leveraged AI to reduce response times for customer service. That was the entire example. It would have been nice to know if they utilized a stand-alone tool or an in-house AI solution. How they trained employees, or how they anchored the data.
Anyway, this is a great entry-level course with three gigantic positives. First, you’re learning from Google staff who are renowned experts. Second, since I’m a visual learner. I enjoyed how they employed simple graphics—such as comparing AI tools and models to a car and the engine. Third, the interactive features were unexpectedly useful. The quizzes and activities actually helped solidify the main concepts. The quizzes weren’t too simple—you had to listen carefully to score 80% and pass.
Final Thoughts
The course offers a handpicked list of AI tools for newbies and includes a glossary of AI terms. Overall, this course is excellent for beginners, visual learners, and anyone who wants to leverage the legit certificate you receive at the end to impress potential employers—or partners.
If you benefited from this article, you might also be interested in watching my overview of Google’s free course on AI—that one’s more theoretical but just as critical. Meanwhile, stay great. Follow for more updates on Tech Updates.
