Some good thoughts on uses and direction.
The AI skills you MUST learn in 2023
Putting you ahead of the curve with real examples
By Ignacio de Gregorio
You won’t be substituted by AI, but you will be substituted by a human that knows how to leverage AI.
Same outcome, but a totally different course of action.
You shouldn’t fear AI by itself, but you should be shitting in your pants if you’ve seen ChatGPT and other Generative AI models and you haven’t started thinking how can you incorporate them into your daily work.
But you can start changing this as soon as… today, as the answer to this issue is known, and goes by the name of Prompting.
A skill so powerful that I wouldn’t be surprised at all if it becomes a standard requirement in job descriptions in almost every industry, and an insanely valuable competitive edge over others if you harness its power.
Embrace or die
Mckinsey, the most prestigious consulting company in the world, sometimes referred directly to as ‘the Firm’, categorized economic activity more than ten years ago into three categories: production, transactions, and interactions.
Finally exploring the unexplored
For production, we all know how machines and robots have completely revolutionized how we generate products, and AI has helped by making these processes more efficient.
With transactions, digitalization, and more specifically, automation with technologies like RPA, are making them a much more streamlined process, with AI enabling the automation of more complex processes.
But until now, interactions were an untouched economic activity for AI, as the results of applying AI in these cases were, well… mild.
Generative AI has completely changed that.
Now, companies are completely confident that AI is going to take over how humans interact with others and with machines, and are ready to fully embrace it.
Thus, if that fact becomes a competitive advantage for your career or a complete burden, will depend ultimately on your capacity to learn this key skill.
This one will hit you close
Let’s see some examples of what I mean:
Many writers, especially newspapers, are already generating a considerable part of their writings using AI and then editing it using humans.
Marketers are generating impressive copy using AI, or leveraging the model’s ample knowledge for idea brainstorming
Companies are already rethinking how ChatGPT can be used for the automation of complex tasks before considered impossible to automate, with action transformers already making an appearance
Companies are also studying ways to how ChatGPT can revolutionize customer service with tailored chatbot solutions
But Generative AI isn’t only impacting how humans perform their day-to-day work, it’s also going to impact how companies operate.
Internet-native companies will become AI-native
Microsoft, by acquiring a majority stake in OpenAI, the company behind ChatGPT or DALL-E, has recently announced the inclusion of ChatGPT into Azure, its cloud computing service.
This is a huge deal, as it allows any company using Microsoft’s cloud services to easily leverage ChatGPT into their systems.
But that isn’t enough for Microsoft.
They also are embedding ChatGPT into its search engine, Bing, and to all its Office365 software suite. Yes, your PowerPoints, Excels, or your Words, among others, will also leverage ChatGPT to enhance your experience.
Needless to say, ordinary people don’t realize what’s coming, and, for that reason, many of them will quickly miss this life-changing opportunity I’m presenting to you today.
Because striving for the continuous reinvention of your skills is a must in the tech-driven world we happen to live in.
You shouldn’t feel tempted to embrace AI, you should be feel forced to do so.
The impact of AI on our daily interactions will be so huge that lots of AI-based jobs will surge as demand for AI experts will soar.
And by experts I don’t mean AI researchers or scientists; simply people that have become experts in applying these technologies.
And that leads you to Prompting.
But what on Earth is this?
Learning to speak to AI
To understand how powerful a skill prompting will be, we need a quick summary of what Generative AI actually is.
AI is a transformer, but not the way you think it is
Generative AI models are models that, when receiving a text prompt, give an output related to that input, be that an image, written text, spectrograms, or others, depending on the use case.
These AI models, at least the most successful ones, fall into the category of Large Language Models, transformer models that use an attention mechanism to work.
In layman’s terms, these models are capable of understanding context from inputs in a much more efficient way using this attention mechanism, which allows them to understand the context and what parts of the input matter the most. This makes these models much more memory-efficient than past models.
But this raises a question. How do we teach context to AI models?
The answer to this question is text embeddings, the hidden power behind ChatGPT and other generative models.
Understanding context using numbers
Text embeddings are vector representations of words. That is, turning a word like ‘human’ into a list of numbers.
By doing this, not only do we transform words into an understandable format for computers, we’re also capable of adding the concepts of direction and similarity to words.
But what do I mean by this?
By making ‘human’ and ‘girl’ have similar vectors — similar numbers — we are essentially teaching machines that those two words, albeit their grammatical differences, share a similar meaning and, thus, are related.
In the contrary, by making ‘cat’ and ‘car’ have different directions and a considerable distance (very different vectors), we are teaching AI models that these words, despite looking similar, will be used in different circumstances and, thus, have low relatedness.
With these vectors you can even add and subtract words, for instance taking ‘human’, substracting ‘male’, adding ‘female’ and getting ‘woman’.
Crazy, huh?
In other words, text embeddings are semantically-meaningful representations of words and sentences in a way that machines can understand and compare them.
Bottom line, it’s a way of using numbers to teach meaning and context to computers.
Text embeddings are so powerful that they are already being researched as potentially optimal to detect dementia, and its most common form, Alzheimer’s disease, from spontaneous speech, detecting patterns that even doctors aren’t noticing.
However though, despite the incredible features that LLMs have, we mustn’t forget that, at their core, they are simply predicting the most probable response to a given output.
Consequently, if they receive bad inputs the responses will also be bad. And, considering you’re charged for every request you make, optimizing the way to communicate with AI is critical.
In other words, for all intents and purposes, how you communicate with them matters. And being proficient in this matter is what Prompting is all about.
“I speak English, Spanish, and… AI”
Prompting represents how humans communicate with AI.
It’s going to be amazingly relevant. A competitive moat to give you an edge over your colleagues. To me, prompting will almost become an art, and for the artists mastering this craft, the possibilities will be endless.
To understand how important prompting is to communicate effectively with these systems, we can see this example I generated using ChatGPT’s public beta:
My first prompt to ChatGPT
As you can see, ChatGPT, despite receiving a fairly simple math riddle, is completely incapable of providing a correct answer.
Now, let’s do this same request with a twist.
My second prompt to ChatGPT
Magically, by simply telling ChatGPT to take a step-by-step approach, much like you would tell your kids when facing this same question, this results in a correct answer.
Again, how you communicate with AI matters.
Now, think about the possibilities this opens for your day-to-day work.
AI as a productivity booster
With Generative AI, once you learn to use it, your productivity will skyrocket. And that’s a huge deal from a business perspective.
Assuming that companies will surely embrace these models, it’s extremely important for you to be capable of rising above your peers by being the ‘go-to’ guy for generating high-quality results from AI.
And that’s what learning how to prompt gives you.
What’s more, I believe the field of Prompting is going to be so huge that companies will start looking for specific ‘Prompting’ jobs in the near future, much like they did with Google search experts.
Time to shine
But these opportunities aren’t going to blossom by default. You have to chase them. Therefore, prompting should be a key skill for you to learn in 2023.
In times of considerable change, those that detect trends before others maximize results. Don’t get asleep at the wheel and, instead of crying that AI is here (nothing you can do about that) start learning how to prompt, and, soon enough, you’ll have a clear edge over 99% of society.
A final word
If you’ve read this article, you’re already ahead of 95% of society when it comes to AI. But that still leaves a lot of people at your level.
What if you’re capable of being above 99% of society? That’s a totally different level.
If that’s you, I have news for you.
I can’t promise you wealth, but I can promise you knowledge, and there’s no wealth without knowledge.
TheTechOasis, the newsletter that simplifies Tech & Crypto
Landing page of The Tech Oasis newsletter about technology, crypto, and markets in a digested manner so that even…
www.thetechoasis.com
Become aware of the technologies that are shaping our future. Be ready to profit from that knowledge.
You also have the option to become a Medium member, opening yourself to millions of articles related to any topic of your interest and, in the meantime, supporting this humble writer.
Join here! Don’t be afraid to level up your game.
Artificial Intelligence, AI Technology Business
Investing
Ignacio de Gregorio Written by Ignacio de Gregorio, 11.4K Followers
AI insights that prepare you for the world that's coming. Opinions are my own. Sign up for my newsletter at https://thetechoasis.beehiiv.com/subscribe ...
No comments:
Post a Comment