Iflexion
ML

Machine Learning: The New ‘Gold Rush’

Scientia potentia est” is a Latin adage that means “knowledge is power”. This phrase is commonly attributed to Sir Francis Bacon and its most common modern interpretation is ‘information is power’. There has never been a time in human history when this phrase was more relevant, as each day humanity creates over 2 Quintillion bytes of data.

This reality has manufactured the big data boom that the world is currently experiencing. All of this data has to be processed, analyzed and stored in some way. The need for effective and transparent data management and processing techniques has paved the way for data science and its growing ecosystem. That’s why ‘data scientist’ was the hottest job opening in 2016.

But data science is only a symptom of the growing ecosystem that’s using the incredibly vast amounts of data that we generate and collect daily. Simply put – data science can’t keep up with the amounts of data that we generate and the applications that we want to utilize it for. That’s why the current pool of information technologies, including data science applications, serve as lubricant for the rapidly expanding niche of machine learning.

Machine learning is the next evolutionary step for data management and processing, as it allows us to amplify the value of collected information. It allows us to teach machines by using data as the metaphorical ‘fuel’ for programs and applications.

This self-learning mechanism allows software to evolve and greatly amplify its own efficiency. This is one of the underlying principles of artificial intelligence. It has gotten to the point that some people already consider software that doesn’t include a self-teaching mechanism, to be legacy software. This is perfect for data-intensive applications, like data security and operational predictions of various sorts.

That’s why machine learning is becoming the ‘gold rush’ of the tech world. Let’s take a look at some of the details surrounding this rush and try to get an idea of exactly why this is happening.

The Software Version of a Swiss Knife

Machine learning has an incredible variety of applications in our data-driven world. In its current state, machine learning has a number of very promising use cases.

ML-1
The % of work activities that can be accomplished within applications using ML. Data source: McKinsey

At the same time, capturing value from data and analytics for a variety of other applications is a problem for current standard data pipelines. This is something machine learning is perfect for – extracting value through optimized algorithms and processes. And there’s a lot of room to grow.

ML-2
Current value extracted for a business niche via analytics in % of its actual potencial. Data source: McKinsey

The labor market is even more ripe for a takeover by machine learning. Not afraid of losing your job to a ‘machine’? Think again.

ML-3
Monetary value of jobs that machine learning can automate as of now, in $ billions. Data source: McKinsey

Jack of All Trades, Master of…Some

There are still applications where machine learning excels better than anything else – for a variety of reasons: some niches have more data available, others have a history of machine learning applications and experiments.

Here are machine learning use cases that resonate the most with current business requirements and the availability of data, which these specific applications require:

ML-4
Top applications for machine learning

Follow the Money

If that’s not enough, just look at the current startup landscape for machine learning/artificial intelligence.

ML-5
Machine learning startups are booming

Machine learning startups are rolling in dough. And, as a result, everything else follows suit, like algorithm marketplaces that have only recently become a thing. Check out how machine learning algorithms stack up against one another at one of the biggest algorithm marketplaces – Algorithmia.

ML-6
The actual number of available algorithms at Algorithmia

Bridging

Machine learning is already taking over our lives. You’ve used your phone’s assistant at least once and if you liked the experience, you’ve used it more than a couple of times. While there are still major gaps between performance and expectations, machine learning is going to change that by refining your experience with phone assistants. With each and every iteration, Siri and Google Assistant will become better and people will learn to accept them.

ML-7
% of people who used their phone’s AI assistant at least once. Data source: Business Insider

Scientia Potentia Est

As we established before – ‘information is power’. That’s why machine learning research is undergoing a very active phase, with global brands building their collective knowledge on the subject of machine learning and its potential applications.

And it’s no wonder, as with current growth projections, machine learning/artificial intelligence products for the enterprise market are going to absolutely explode within the next 8-10 years.

ML-8
Growth projections for AI/Machine Learning in $ billions. Data source: Tractica

That’s a 50x growth in nine years. The actual California ‘gold rush’ didn’t even last that long and certainly wasn’t anywhere close to being that lucrative.

There’s More Where That Came From

As we established before, big data is the main ‘protein’ in machine learning’s diet. Big data related services, like data management products and infrastructure that can actually support the data, are growing at an astonishing pace with projections that dwarf even machine learning.

ML-9
Big data related spending in $ billions. Data source: A.T. Kearney

This growth is what’s pulling machine learning into the orbit of various high profile businesses and raises the overall interest towards machine learning and AI technologies. For example, Google recently acquired Kaggle – the biggest data science platform – you know things are about to get serious when Google is joining the game. JPMorgan is also catching up with their machine learning initiative. IBM also has initiatives that are different from Watson, but are also business-oriented.

This booming interest towards AI tech also leads to a new phenomenon in the business world. As if you didn’t have enough acronyms like SaaS and PaaS to learn, check out MLaaS: Machine Learning as a Service, a niche that’s exploding, for a lack of a better word.

ML-10
MLaaS Market Forecast, 2016-2021 in $ millions. Data source: MarketsandMarkets Research

Modern ‘Gold Diggers’

The labor market is a great indication of tech trends, as it instantly adjusts to the demand. And even data scientists are making fun of how machine learning is taking over the labor market. Laughter through tears, as they say.

cartoon-data-scientist-sexy-robot

However, it’s still obvious that machine learning products can’t completely fulfill the demand. That’s why the demand for R and Python developers is growing on the labor market. These languages are the basic tools for data science and it’s obvious that they’re still in high demand, at least locally.

Conclusion

It’s a fight or flight situation for many businesses: if they don’t jump on the machine learning bandwagon, they risk getting left behind with data scientists taking months to realize and analyze algorithms, while machine learning can do that in a couple of hours. The same goes for software developers and product companies that are not currently employing machine learning in their products – sooner or later this has to happen. Machine learning is one of the biggest AI niches that’s actively growing because it actually delivers results through better business predictions and decision-making.

John Barnett

  • Aria Haneul

    I’d read the NPR article you linked to when it first came out, and I agree. It’s hard to envy the position data scientists are in now considering the immense amount of data being generated daily. I’d love if you guys did an article about how technology can help the Cancer Moonshot or similar projects. I’m learning so much about AI and machine learning from your articles. Keep it up!