With AI engineering jobs ranging from $175,000 to over $300,000 a year, here’s an attractive pursuit. But to be a good AI engineer, it’s not enough to know how to ask questions and yell it on LinkedIn. You need to combine the disciplines of AI, programming, language, problem solving, and even art to thrive in this career path.
Prompt engineering consists of creating interactions with generative AI tools. These interactions can be conversational, as you’ve no doubt seen (and used) with ChatGPT. But they can also be programmatic, with (prompt) prompts embedded in the code, the rough equivalent of API calls. Except you’re not just calling a routine in a library, you’re using a routine in a library to talk to a large language model.
Before we talk about the specific skills that will come in handy to land that speedy engineer job, let’s talk about one trait you’ll need to make it all work: willingness to learn. Although AI has been around for decades, the explosion in demand for generative AI skills is recent. The field is evolving very rapidly, with new breakthroughs, new products, techniques and approaches constantly appearing.
To keep up, you have to be more than eager to learn – you must be voracious in learning, searching, studying and absorbing everything you can find. If you keep learning, you will be ready to grow in this career.
So here are the six skills we recommend you acquire to become a speedy AI engineer.
1. Understand AI, ML and NLP
For starters, understanding how artificial intelligence, machine learning, and natural language processing work is essential. If you intend to interact with large language models, you need to understand the different types of LLMs that exist, the types of tasks LLMs do well, and areas where they are unreliable.
This doesn’t necessarily mean you have to become a computer scientist who can create your own LLM, but it does mean you have to understand the inner workings and capabilities of the tools around which you’re trying to build a career. To do this, you will need to train yourself by all available means, including traditional courses, reading numerous articles and technical documents, attending conferences and carrying out your own experiments.
2. Clearly define problem statements and specify detailed queries
Basically, this skill is the ability to communicate with clarity. Prompt engineering is all about telling the AI what you need. To do this, you need to be clear about what you want to get out of the interaction.
Here is an example. Suppose you want to know more about Valence, the prefecture of Drôme. You must be clear on at least two points. First, you need to explain the kinds of things you want to know, whether it’s the local political structure, city management issues, traffic, or where the best bakery is. Second, you need to be able to tell the AI that you’re talking about Valence in Drôme, not Valence in Spain.
You will also need to learn how to explain how to set expectations for AI, how to position it to understand the perspective it needs to use to deliver value, and the context and scope of the problem you want it to deliver. it resolves in a given query.
Here too, you will need to understand the limitations of the different LLMs and know how to work around them. For example, if you want to write a white paper, you might need to generate an outline first and then ask the LLM to write each section separately. Also, remember that a clear prompt doesn’t necessarily mean it’s short. Longer questions can lead to more specific and relevant answers.
The conclusion is simple: favor clarity and make sure you are able to communicate without assuming that you understand.
3. Be creative and develop your conversational skills
Prompt engineering is much more a collaborative conversation than a programming exercise. Although LLMs are not sensitive, they often communicate in a manner similar to how you would with a colleague or subordinate.
When defining your problem statements and queries, you often have to think outside the box. The image you have in mind may not match the internal AI representation. You will need to be able to consider a variety of different conversational approaches and stratagems to achieve the desired results.
If you want to become a nimble engineer, the experience of debating, negotiating and even selling will serve you well because it will allow you to practice the skills of conversation, persuasion and collaboration that are so critical to getting the desired results from generative AI systems.
4. Get to know writing styles and artistic styles, and develop your expertise in the field
Not only do chatbots write answers for you, they often do so in the style you request. For example, I asked ChatGPT to write in the style of Jony Ive, whose excessively flowery descriptions of Apple products have become legendary.
And that’s not all. You can also use styles for image generators like Midjourney. You can create images in a cinematic style, in the style of 1940s cartoons, and in a wide range of photographers.
For example, I used my portrait photo on Facebook and sent it to Midjourney with the keyword “cubism”.
Here is the original image I sent to Midjourney. As you’ll see, she’s a little confused by the microphone, but she’s still interesting. David Gewirtz/ZDNET
In this case, I kept the prompt minimal, but used what is called a “base image”, which is my Facebook icon. Here are four variations of the “cubism” style. Which do you think resembles the original the most? Midjourney and David Gewirtz/ZDNET
In addition to understanding writing and art styles, it is important that you develop (or be able to access) expertise in the area for which you are creating prompts. For example, if you are working on an AI application for automotive diagnostics, it is important that you are familiar enough with the field to be able to get the answers you need and understand if they are correct or not.
And here’s a skill within the skill: get to know the extensions and plugins used by your favorite generative AI tools. Over time, these extensions and plugins will help you do things you can’t do with the standard AI tool. So learning and using the add-ons will allow you to not only keep your skills up to date, but also accomplish things you couldn’t otherwise.
5. Develop scripting and programming skills
It goes without saying that programming skills would be useful. While there are a few engineering jobs that simply interact with chatbots, the highest-paying jobs are likely to be integrating AI messages into apps and software that then provide unique value.
Although you’re not necessarily expected to write the full application code, you’ll bring much more value if you can write some code, test your prompts in the context of the applications than you build, run debug code, and generally be part of the interactive programming process. It will be much easier for a team to move forward if prompt engineering is an integral part of the process, rather than having to add and test it as a completely separate operation.
Also, coding skills are valuable on their own. Everyone should have some basic coding experience.
6. Develop your patience (and your sense of humor)
I firmly believe that it is much easier to be patient if you have a sense of humor. These generative AIs definitely require patience. They will misinterpret requests. They’ll lose track of a conversation the moment you’re about to find something. They can make up answers that are just rubbish.
If you can’t laugh about it, you’re going to have a bad time.
The same goes for programming. Every programmer needs patience. One of the biggest challenges some of my students had when they first started programming was that they couldn’t accept that their code didn’t work the first time it was run. Those who couldn’t hold on didn’t complete the course. In contrast, even those who weren’t very inspired coders, but had the patience to try, fail, research, and try again, did very well.
Think of it this way. AI is a mixture of working with an incredibly literal computer, a learning model that interprets things in unpredictable ways, human team members (some of whom are even more literal than machines), and random nature. unpredictable of the universe.
Patience is not just a virtue. It’s a super power.
Some additional thoughts
There it’s done. I’ve outlined six skills/characteristics you need to be successful as a speedy engineer. But keep in mind that these are only rough guidelines, and it’s a very individual path that you will have to follow.
If I can give you one essential piece of advice, it’s this: DIY!
Choose projects that interest you and build something. Team up with a few friends and see what you can produce. Hands-on experience will get you much further than a list grabbed from the Internet.
Get out there and do some speedy engineering. Create small applications. Take classes. Build things. Then not only will you no longer be someone who wants to get into speedy engineering, but you will be someone who has done it and has something to show for it.
Try it, it’s the only way to know.
Warning : Use of AI-generated images may result in copyright violations. Caution should therefore be exercised if the images are used for commercial purposes.
Source: “ZDNet.com”