Welcome to 2024: The Year Where AI is No Longer an Option

Why everyone should learn about machine learning

A huge welcome to the many new members of Society's Backend! Thank you for joining our network where we explore the complexities behind machine learning and the software that keeps it running. Changes are coming soon to make Society's Backend more interactive, supportive, and collaborative for all members. Stay tuned!

This month I'll be going over numerous machine learning resources to review, clarify, and organize the information. I'll be creating a machine learning walkthrough with both technical and less-technical paths. The goal is to make an easy way for anyone to learn about machine learning and its implications. Follow me on X and join me here as I share what I find most important from each resource.

---

As I started going through these resources, immediately something important stuck out to me. Andrew Ng (of Stanford and Coursera) made a very accurate comparison: "AI is the new electricity". Electricity transformed every major industry and, in turn, changed the world. Machine learning will do the same. Similar to electricity, everyone will be using AI and everyone will need at least a baseline understanding of how it works.

This got me to thinking about how we can share information about AI and machine learning with non-technical consumers so they can understand its implications and how it affects their life. I asked myself: What resource would I give my parents to understand everything they need to know about machine learning? I realized I didn't have one.

This reinforced a thought I've had for a long time: the single biggest issue with machine learning isn't data or compute, but accessibility. Machine learning is nowhere near as accessible as it needs to be. We have many high-quality resources that teach machine learning concepts (for free!) but navigating these resources is like navigating the ocean without any sort of guidance all while the ocean is growing larger at the same time. The field is so vast and growing so quickly we need a resource just to navigate learning resources.

While there are ample resources for learning the technical aspects of machine learning, we lack material on non-technical learning that consumers need. These are the resources someone with minimal technical knowledge (like your parents) might need. There are a few newsletters that do a good job of this, but those resources aren't enough.

This is part of my goal with Society’s Backend: create a network of those studying machine learning from all levels of experience so we can understand artificial intelligence together. This is also why the machine learning walkthrough I’m putting together will include both a technical and less-technical path for learning.

You might argue that consumers don't need to comprehend machine learning and can rely on those who do. However, I liken it to understanding how the internet works. Knowing how companies target ads, how online communication functions, and how scammers exploit internet users is essential to navigate the digital world. The only difference is consumers may need to understand more about machine learning.

Let’s take prompt engineering as a specific example. Many non-technical individuals use generative AI and it’s likely that all will at some point. While a technical understanding of the machine learning behind generation isn’t necessary, everyone using generative AI will have to understanding prompting to ensure proper generation. Understanding prompting requires a baseline understanding of AI.

This is because it's not only about consumers using machine learning, but also because artificial intelligence has implications outside of just technology. It is changing:

• Ethics: Look at the NYT vs ChatGPT lawsuit. This will change how we view copyright law.

• Economics: AI is both a cause for concern regarding job displacement and a driving force behind economic growth.

• Environmental: AI can help address environmental issues but it also creates some of its own. How do we balance the problem-solving of AI will its power-hungry nature?

• Education: LLMs (large language models) have proven to be very effective educational tools. How will the future of education be impacted by LMMs (large multi-modal models) that can generate text, video, and audio, especially after future performance increases?

• Social: The advent of AI chatbots and their usage in scenarios like AI girlfriends raise concerns about the future of human social interactions.

In addition to these areas, machine learning will significantly affect various industries. AI advancements will enable faster research and development in every other industry. For instance, AlphaFold has revolutionized medical research by enabling rapid protein exploration via machine learning. It turned a day- or week-long task into one that requires only minutes.

Imagine bringing that same type of efficiency and accuracy gain to tasks like surgery or big data analysis. Tasks that require a specific skillset and a lot of time to accomplish. Even common tasks like video and photo editing and transcription/translation will take much less time.

In conclusion, understanding artificial intelligence isn't an option anymore—it's a necessity. As research in machine learning progresses, we need bright minds from all industries to ensure it is trustworthy and its development benefits humanity.

The bottom line: Machine learning has gone from being A solution to problems, to THE solution. It's advancing rapidly and this is just the beginning.

---

Join Society’s Backend for articles like this. Learn about the engineering behind machine learning systems, the most effective way to learn about AI, and how to use AI in the products and projects you build.