Machine Learning And The Hard Problems

The ground is moving under the Machine Learning field. There are so many factors aligning behind our domain at the same time. I have been in technology for 27 years, and this is an anomaly.

The WEF event last week had substance behind the technical talks. The heads of major funds started using a term that I have been evangelizing for some time, “technical strategist.” It began with Jim Breyer calling Satya Nadella and Mark Benioff technical strategists. He said, “We are investing in technical strategists.”

That seemed to open the flood gates. The prospect of a downturn has moved sentiment from speculation to safety and quality. The focus is applications of machine learning targeting high-value opportunities. The field is expected to solve the hard problems instead of making incremental improvements.

Senior leaders have told a story about leveraging data and AI. In the past, the story was enough. When everything grows, no one cares if the growth lines up with the story. Growth is hard to find and will not get any easier for the next 12-18 months. Not all businesses are growing anymore, so the story is not enough.

Investors are looking for businesses that can leverage technology to solve internal challenges and deliver value to customers in new ways. Here is how companies can establish credibility.

The Team

Individuals do not solve hard problems. The unicorn and solo rockstars are things of the past. Hard problems are also not solved through research papers with no practical applications. SoTA is a benchmark of the past.

The flight to safety and solutions will increase demand for 3 types of people in our field. The first is people who can find hard problems that machine learning is uniquely capable of solving. The second is people who can bring solutions to the marketplace.

The technical strategist is just the first piece because strategy without execution is the slowest path to victory. Slow is not tolerated in the current environment. The second critical piece is the team.

A credible Data Science team handles 4 phases.

• Data

• Applied Research

• Model Development

• Deployment and Support

A high-performing team with top talent requires the third type of person, capable leaders. They need to operate at multiple levels, from managers to C-level executives.

According to most surveys, 2/3s of businesses are stuck in the prototyping and exploration phase of machine learning maturity. A significant share of experienced Data Professionals has never delivered simple solutions to production. Fewer have delivered solutions to hard problems. Many leaders have not overseen a complete team or been accountable for delivering solutions.

The talent pool is already too small to support demand. With higher expectations, the qualified talent pool is even smaller. Assembling the team is obviously expensive but what is overlooked is the process of building the team is knowledge-intensive. The build-out process is complex.

Data and Analytics organizational development requires a deep understanding of each role to create job descriptions and assess talent. Machine Learning tools and infrastructure must be built out in parallel to the team, so everything is in place for them to execute. The business needs training and transformation that also aligns with the organizational ramp-up and product roadmap.

This is the team investors are looking for, whether at a large company or startup. A business talking about adopting data, analytics, and models must have that team in place, or no one will believe in their story. Safety and quality avoid the visionary in favor of the pragmatist.

The Blueprint And Roadmap

An amazing team working in a silo is set up to fail. Technology driving business strategy and prioritization is also a set up to fail. Solving hard problems is a business-wide effort. It must be driven by strategy and senior leadership. Every business unit needs a high level of strategy awareness and goal alignment. The business and operating models used to be enough, but modern companies need a third strategic model.

Solving hard problems starts with a Technology Model. This explains why the business uses technology, how each technology generates value, and how the business can deliver technical solutions and generate revenue in the face of competition.

The Technology Model makes it possible to define the hard problems the technology and business should solve. Boxes are checked to ensure the business is identifying, validating, and prioritizing appropriately.

There is a high-value problem that machine learning solves better than alternative technologies.

The business has competitive advantages around the problem, customers, and solution.

OR

The business will create a competitive advantage around the solution.

The team is capable of building the solution and delivering it to production.

 The business is capable of delivering the solution to customers or users and generating returns.

The roadmap is a repeatable process to go through 3 phases: productization, commercialization, and monetization. When investors dig into the business, the blueprint and roadmap establish credibility just as much as the team does.

Outsourcing The Process

My business and the few others in my space are seeing a lot of new interest. Companies are looking for ways to accelerate their transformation and build-out. Bringing in external consultants is usually the preferred approach.

Accenture, Boston Consulting Group, and Deloitte are the 3 big players. Multi-billion dollar businesses are turning to them in growing numbers. Their services are expensive, but they bring a small army to engagements. Large companies are sitting on piles of cash, so they are willing and able to pay for growth.

The number of companies looking to bring those 3 companies on board is more than even the big consulting houses can handle. It is crazy to think about those 3 not having enough people to meet the demand, but they don’t. All 3 are ramping up internal training and hiring to support the growing business. Small players like me are picking up the overflow and providing training support simultaneously.

Some companies are still unserved. Data and AI Strategy are both emerging fields, so there is not enough experience to go around. How many more people do we need? My best guess is 25,000, maybe a little more to service the market.

Large businesses need 5 to 10 people each, and there are about 1000 who are underserved or unserved all together. My market research puts the number of SMEs investing in future growth around 50,000. They rarely need a full-time consultant, so 1 consultant can usually service 2 to 3 businesses simultaneously. They also do not need people for as long as the larger businesses do. Engagements are 6-12 months, depending on the size of the business.

There will be a disparity between those who can outsource the strategy side and those left underserved. The most likely winners are larger companies with higher budgets. That is not good for the long-term growth picture.

The Big Picture

SMEs will be forced out because they cannot access talent to transform and staff their teams. A coming wave of consolidation will exacerbate the problem.

Knowledge will be concentrated even further into the hands of a small number of very large companies. That decreases competition and slows growth rates. Large companies will quickly pull ahead so far that it becomes difficult to remain competitive or enter the marketplace. The advantage of startups and SMEs is their speed and agility. Put up the barriers of high entry costs and significant competitive advantages, and startups will have difficulty finding an entry point.

Startups are needed to force innovation and increase the pace of progress. SMEs power much of the middle-class job growth and provide spending power to those consumers. If both groups take a significant hit, that is terrible news for the global economy.

You will hear a growing fear and sense of urgency coming from large businesses’ CEOs. Everyone has had time to read the tea leaves from Davos. The central theme behind everything was new technology. If access to new technology is concentrated into the hands of a few, CEOs want to make sure they are one of the few.

In 12 months, we will have to deal with the longer-term issues of losing SMEs. We are on the clock. The business world needs talent, and you are the people who can help train them. If your company does not have an internal training and upskilling program, it is time to build it. As top talent is laid off, jump at the opportunity.

Apple invested into the last downturn. The knee-jerk reaction now is to cut everything back. Apple emerged from the Great Recession as a powerhouse. Targeted investments now will yield dividends in the long term and are critical to survival.

I find myself saying this to Data Professionals a lot. I know you did not sign up for this, but it is you or no one. I know this is not the reality you would have chosen, but this is our reality. It is not the opportunity you asked for, but it is a huge opportunity, nonetheless.