×

Once registered, you can:

  • - Read additional free articles each month
  • - Comment on articles and featured creative work
  • - Get our curated newsletters delivered to your inbox

By registering you agree to our privacy policy, terms & conditions and to receive occasional emails from Ad Age. You may unsubscribe at any time.

Are you a print subscriber? Activate your account.

Machine Learning and AI: When to Start?

By Published on .

Today is a good day to start thinking about investing in machine learning tools for your business, writes Shelly Palmer. But taking some areas slow could help speed you up in the long run.
Today is a good day to start thinking about investing in machine learning tools for your business, writes Shelly Palmer. But taking some areas slow could help speed you up in the long run. Credit: Michael Short/Bloomberg

If you want to build a ship to take humans to Proxima Centauri (the nearest star to the Earth), when should you start the project? If you start today, you might be ready to launch your ship in about 500 years, and accounting for exponential technological advances, you might get there in 10,000 years or so. (It's 4.243 light-years away; that's 24 trillion miles.) However, if you wait 5,000 years to start building your ship, you may only need 500 years of travel time. So waiting 5,000 years to start the project might get you there 5,000 years before people who start the project today. This completely hypothetical thought-starter is one of my favorite ways to explore investment strategies in the age of exponentialism.

When to start using machine learning in your business is not a hypothetical question; it's a question you must answer today. Not because I say so, but because your competitors are working on their answers as you are reading this. So here are a few thought-starters to help you explore your machine learning investment strategy.

Are You Ready?

To begin an intellectually honest discussion about value creation using machine learning systems, you will need to assess your organization's data maturity as well as its readiness to accomplish its data-driven goals. You should do an audit of your data governance, data warehousing, data scientific research capabilities, and data hygiene, and you should take a close look at the sources, uses, volume, and veracity of your first-, second-, and third-party data.

Play, Experiment, Explore

Amazon, Google, Microsoft, and many other big-name tech companies have suites of machine learning tools that are wonderful places to start your journey. If you want to have some fun, go to Google's AI Experiments page and just play. If you want to get a bit deeper into practical experimentation, check out Amazon AI. Monkey Learn has a nice add-on for Google Sheets that is easy to use and requires absolutely no programming skills at all. The more you play with the technology, the better you will understand machine learning's potential to significantly increase productivity.

Commoditized Machine Learning

Salesforce recently announced, "New AI Breakthrough from Salesforce Research Boosts Productivity with Text Summarization." I do not work for Salesforce and I don't have a dog in the hunt, so I'm not advocating for this particular tool set. That said, everyone who is willing to pay for this capability is going to have it. You don't have to train it yourself, and you don't have to invent it in-house. You just have to subscribe to it. This is only one of thousands of such services that are going to appear in the next year or so. Set up an in-house system to ensure that you are aware of all of the commoditized machine learning productivity tools that are likely to impact your business. More importantly, think through how many machine learning systems you will not need to create for yourself.

Machine Learning Systems Are Trained

Exponential improvements in machine learning are pointing to a future where systems are trained, not coded. Today, software engineers write programs (code) to instruct computers to perform tasks. Those days are clearly numbered. Sooner rather than later, computer programming will evolve into something more akin to animal training. As the systems evolve, the training will closely resemble parenting, and ultimately, machines will train machines. I have no idea what will happen after that.

Worker Displacement and Replacement

As you experiment with machine learning tools, you will quickly come to the realization that some workers will be replaced by this technology and others will be displaced. The ramifications are not well understood. The future of work will be dramatically impacted by machines. You must use your own experience, statistics, and common sense to set a timeline you believe in. You will be laying off some white-collar workers and retraining others, and so will every other company. It is going to be a major policy issue in the very near future. Remember, this is the age of exponentialism.

When Do You Start Seriously Investing in Machine Learning Tools for Your Business?

Today is a good day to start thinking about investing in machine learning tools for your business. Commission the data audit ASAP. Commit an appropriate amount of resources to best-in-class data governance. It is impossible to turn data into action unless you have easy access to it. Then, subscribe to the commoditized machine learning services wherever they make sense. The productivity increase will create value.

As for building and training an in-house machine learning system? Let's get together and talk about slowing down to speed up. Proxima Centauri is just a few light-years away.

Most Popular
In this article: