Should You Lead or Follow When It Comes to AI?

More and more businesses are incorporating AI into their companies, but the majority of companies have yet to explore this powerful technology. Larger companies like Google, Facebook, IBM, and Amazon are leading the way. If you’r are a key strategic decision maker of a company, the question you’re asking yourself is should you lead or should you follow? In order to answer that question, let’s get some background first on the state of AI. Secondly, I’ll talk about the pros and cons of waiting versus diving in. By the end of this article you should be closer at arriving to your decision.

Where are we on the S-curve?

In order to understand when to invest in a new technology, we must first ask, where are we on the S-curve of AI technology development and adoption? (Exhibit 1) The time to invest is sometime in the first half of the curve before the growth switches direction, and then start diminishing investment in the latter half.

Exhibit 1

Although it is hard to pinpoint where exactly we are on the curve, we can make an educated guess based on the # of AI papers being published (Exhibit 2), the growth in applications of AI, and the achievements in the complexity of AI. We also know that AI relies on data. Data initiatives didn’t start to really take off until the past 10 years or so. Since we are still perfecting data collection and codification, we know we have not yet reached the peak of AI and are probably nowhere near it. But we also know that based on the recent uphill gains in adoption and application in the past 2 years, we have definitely lift-off the runway. Now is a time where companies are starting to question, is now the time to invest in AI or should I wait a little longer? First, the progress towards AI is speeding up, not slowing, as more and more companies are adopting and working on it. If you do decide to wait, it won’t be much longer now before the curve gets too far ahead of you. So what I am saying is if you decide to not invest in AI yet, then you should be on your toes, watching the AI landscape intensely for your moment.

Exhibit 2

To Lead or Not to Lead?


If you don’t have a spare a few $100K

Cognitive computing, although cheaper than it once was, is still expensive. Right now one of the cheapest ways to do this is through AI chips which can cost you around $10K each and will likely need around 8 to be competitive in the amount your are computing. Additionally you will need AI talent, which can cost you between $130-200k per head annually. If you don’t have this financial resource then you might be a follower. In time the cost of training will come down.

If you don’t have a collaborative culture

Research shows AI projects are most successful in a collaborative environment. Would you consider your organization collaborative?

How can you make up lost time?

Hack & Train faster

The advantage of being a follower is that you can potentially use open source machines and algorithms to train your own AI in what is called adversarial AI where machine teaches machine. This will shorten your training curve tremendously.


If resource-endowed

Leading will involve a significant commitment of financial resources and R&D capabilities. If you have these then you might be a good candidate to lead. Right now mostly big companies are leading the way because they have these kinds of resources.

If you are close, but need an alternative

But if you are a smaller company and don’t have these things, then you could get creative and consider partnerships to reduce the financial risks or to provide you R&D capabilities.

If you have an innovative culture

You also want to think about your strategy and what kind of company you want to be. Is innovation critical to staying abreast of a competitive environment? Is being seen as innovative critical to winning over customers? If so you might want to take a leadership role.

How can you defend your position?

Anti-Hack Plan

Be weary. Although it may be difficult to reverse engineer an algorithm, it is possible smaller competitors might use your algorithm to quickly train theirs to get it up to speed. Also be weary of runner-up competitors trying to take down your AI by feeding it garbage data. You will have to make further investments to mitigate against this.

The 2 critical factors that make your AI successful

One thing to remember is that the greater variety of data a firm has to inform their AI the more powerful it will be. Therefore it is critical that you start collecting data early on, even before you think you need it. Once others have a headstart it is harder to catch up. Case in point, look at how Google surpasses everyone and has minimal competitive threats. No one can possibly take on Google. Leaders who dive in early enough will earn this competitive advantage and make it harder for others to follow.

Customers have shown that they have a low tolerance for AI that doesn’t work. Therefore you can expect that in order for customers to adopt your AI solution, it will need to be close to 99% accurate. Accuracy wins when it comes to AI, and if your solution is 2nd best and not above this threshold then no one is going to buy your solution. 1st place wins, so be sure that you can get the resources to go all the way.

What should being a follower really mean?

If you decide to follow, this does not mean checking-out and revisiting in 1 year. You should be constantly surveying the AI landscape and educating yourself so that you are ready to jump when the time comes. The curve won’t slow down for you to hop on, remember that, so staying up to speed will help your rapid entry. A good rule of thumb to keep in mind is the later you enter the more rapid your entry must become. If your organization is slow at getting anything done then perhaps you should start sooner than later. One of the ways I recommend companies to enter this stage is to start a mandate from the top, encouraging employees to start exploring this space and learning as much as they can together, like a single cohort. Soon they will be coming back with ideas from the courses they take and the conferences they attend that will eventually trigger the AI idea you will soon be working on. You can help them to start to build their knowledge, skills, confidence, and capabilities with competitions such as the AWS Deep Racer Challenge as a great entry point.

Disruption is disruptive for a reason

Whether you choose to become a leader or a follower in AI, just understand that AI is a disruptive technology that will change how we do business, how we interact with customers, the economy, how we live life and more than can be imagined here. Keeping in this in mind, I just want to remind you of the history we have seen with disruptive technologies, i.e.) Digital photography putting Kodak out of business and Uber putting taxi cabs out of business. Need I say more?

If you would like to learn more about the AI landscape I highly recommend the AI Index Report by Stanford.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.