Why Machine Learning will Dominate Software

This article is originally published at https://www.sharpsightlabs.com

AI is turning the tech world upside-down right now.

With the release of large-language systems like Chat-GPT and various coding copilots, many people are questioning the value of a tech career.

I have good news and bad news.

The bad news is that if you’re just a plain old software developer, AI might undercut your ability to earn large amounts of money.

The good news, is that if you know machine learning, you’ll probably be able to make bank.

Let me explain.

This Week In Startups on the Impact of AI

I’m a regular listener of the This Week in Startups podcast.

It’s a show about (as you might guess) startups, and also the technology industry more broadly.

The show is hosted by a long-time Silicon Valley insider named Jason Calacanis, and he regularly brings on high profile founders, investors, and entrepreneurs.

Recently, Calacanis started doing regular “AI Roundtable” episodes with two other tech insiders: Sunny Madra and Vinny Lingham.

In these roundtable episodes, they’ve been talking extensively about specific AI tools, like Chat-GPT, and others.

But they’ve also been talking about the impact of AI on the Tech industry.

And in their most recent AI-focused episode, Lingham made an important point about the relative value of traditional software vs machine learning going forward.

Software Down , AI Up

Lingham noted that traditional software is likely become less and less valuable (i.e., less able to generate profit).

But he continued by pointing out that the real source of profit going forward will be the AI features on top of that traditional software.

What he literary said was “software products [are] going to zero … You’re going to be paying for the AI features on top of that … You won’t be able to charge for the base product anymore. The money will go into AI.”

I think this is an astute prediction.

To understand why, let’s quickly discuss why this will probably be true.

Money Goes to the Scarcest Inputs of Production

The important thing to remember is that in a free market economy, “the the biggest premiums go to the scarcest inputs needed for production” (Quote: Brynjolfsson and McAfee in The Second Machine Age).

AI copilots and generative coding systems will make it much easier to write code. Code itself is now easier to write and produce at scale. It is now less “scarce.”

So, we’re going to get a lot more software, but the cost of that software will decrease, and the profits of traditional software will probably be eaten away by competition.

So what will be the scarce input of production in technology companies?

Machine learning.

As Lingham noted, AI features will drive real value for technology and software products going forward.

And (as we’ll discuss more in a moment), modern AI really means machine learning.

Now I do understand: AI code-writing systems like Chat-GPT will make it easier to write machine learning code.

But there’s a lot more to ML than just the code.

With machine learning systems, there’s getting data, analyzing data, cleaning data, selecting features, tuning hyperparameters, and a lot more.

Is code part of these tasks?

Yes.

But often, you need an underlying conceptual understanding to know how to apply the code in order to do these things the right way.

Said differently, to use AI coding systems properly in the context of machine learning, we’ll need more people with the right knowledge and conceptual understanding.

AI is moving fast

The other important thing to consider is that AI is moving fast.

There are literally new breakthroughs in machine learning and AI every month.

It will be hard to keep up, but people who are at the cutting edge of machine learning will likely have a huge advantage, simply because skills in those breakthrough areas will be rare by definition. And, after every AI breakthrough, it will take new copilot systems a while to catch up.

Money will flow to the people and companies that can stay close to the cutting edge of ML.

You Need To Learn AI

What does this mean for you?

You need to learn AI.

If you’re in college and studying a tech-related field? You probably need to know a little about AI.

If you’re a software developer, guess what. Your skill will become more and more of a commodity if it’s the only thing you can do. But adding AI skills on top of an existing software development skillset will be very powerful

You need to Learn Machine Learning

Now, let me be more specific.

You need to learn machine learning.

Once opon a time, AI could have meant something else. There were (and really still are) other types of AI outside of machine learning.

But right now, machine learning is what matters.

GPT is essentially a very advanced machine learning system.

All of the generative AI systems (like Stable Diffusion, etc) are machine learning systems.

Going forward, if you want to compete in SaaS and the software industry, you need to know machine learning.

How to Get Started

I’ve been banging this drum for well over 5 or 6 years, but I’ll repeat myself:

If you want to master machine learning, you need to start with foundational data science skills.

Learn Data Science Essentials

That means, if you want to master machine learning, you should start by studying:

• data wrangling and data cleaning
• data visualization
• data analysis

I get it.

You probably think you’re a special snowflake.

You might think that you can just jump to the coolest stuff first (i.e., jump right into deep learning and advanced ML).

No.

That’s what the wannabes do.

People who are true professionals focus on foundations first. I’ve been saying this for years, but people like Michael Jordan, Kobe Bryant, Navy SEALs and many elite performers focus on foundations.

If you want to get really good at creating AI and ML systems, you really need to know some of the foundational data science skills first.

Skipping this step would be extremely foolish, and will hinder (if not ruin) your progress.

Master. the. foundations.

Learn Machine Learning Essentials

After you learn foundational data science skills, then you can move on to machine learning.

For the essentials, you’ll need to learn:

• regression vs classification
• different algorithm types (trees, linear models, neural networks)
• model evaluation and cross validation
• data visualization for ML
• how to do data preparation for ML
• and more

These are foundational ML skills that you’ll need first.

Once you know these machine learning fundamentals, you’ll be able to move on to more advanced topics like deep learning, large language models, etc.

Leave Your Comments Below

So do you agree with me (and with This Week In Startups) that AI/ML will take over software?

Do you see things differently?

I want to hear from you.

Leave your comments in the comments section below.

Thanks for visiting r-craft.org
This article is originally published at https://www.sharpsightlabs.com
Please visit source website for post related comments.