Kenneth Cukier is a data analytics wizard, New York Times bestselling author, and seasoned journalist who oversees analytics and digital product development for The Economist. He’s also a regular contributor to publications like the Financial Times and a fellow at Oxford’s Said Business School, where he researches Artificial Intelligence (AI).
Recently, he visited Red Ventures to talk about big data’s big future in an increasingly digital world. Here’s what we covered:
- What the @#$%! is AI?!
-
How can we use data to
read mindspredict human behavior? - Which industries will big data revolutionize first?
- What steps should companies take to start using big data?
- How can we prepare for the impending AI apocalypse?
Read the full transcript of my interview with Kenneth Cukier below:
Nicole: Copywriter here. I work with words, not data. How would you explain AI to the average Joe? (Or, in this case, the average Nicole…)
Kenneth Cukier: It’s all about spotting patterns – patterns that may not be obvious to human beings, but that a computer can spot easily. A computer has a perfect memory, and it can process hundreds of thousands of questions, transactions, or instructions in a fraction of a second. As a result, a computer can identify things that humans can’t. So whether it’s looking across an entire population to spot traits that predict cancer, or whether it’s a self-driving car, the principle of “AI” and the technique of “machine learning” comes down to spotting patterns and classifying data.
So whether it’s looking across an entire population to spot traits that predict cancer, or whether it’s a self-driving car, the principle of “AI” and the technique of “machine learning” comes down to spotting patterns and classifying data.
N: I consider humans to be pretty unpredictable – are you saying we’re more pattern-oriented than we’d like to think?
K: Sorry to break it to you, but you are completely predictable. Human beings are completely predictable. And the degree to which you’re not predictable is factored into our model. With just a few features, data helps me know 80% of everything about you. For example, I can tell you where you’re going to be at this time next Monday. I could use data from your mobile phone carrier to determine where you’ve been every other Monday, and make a prediction with some degree of certainty that you’re going to be there next week as well. And if I’m wrong because you have a very chaotic schedule, I’d be able to predict that, too.
We like to think that we have free will – and we do to an extent – but from a data perspective, humans are completely predictable.
Another example: the fact that you’re a woman, you’re White, and you live in this area is enough to tell me which party you voted for in the last election. If you told me your religious affiliation, I would know with even more certainty. We like to think that we have free will – and we do to an extent – but from a data perspective, humans are completely predictable.
N: Terrifying, but as a data-driven Creative in the marketing world… I can’t really say I’m surprised. How about jobs? Which industries do you think will be disrupted by big data and AI first?
K: I think Healthcare is where we’ll see the biggest gains, because we already collect the data. It’s just not being used yet. When we start using that data, we can learn – at a population scale – things about the pattern of disease that we would never have known before. But more importantly, we can drill down and look at individual data, and then tailor our diagnostics and therapies around an individual person – rather than the fiction of the “average man” (to whom no single person actually belongs).
Another one: Education. If you stop treating the data of student performance as a stock, and instead see it as a process – something that happens continuously – you can customize teaching methods to fit each person. You may find that one student is learning her lessons far better when she does her reading in the morning. But for another student, it may be the opposite. Maybe he has this golden moment at 8pm because his younger sibling finally goes to sleep and the house is quiet. Who knows? It doesn’t matter. You won’t catch the causality, but you’ll get the correlation. And when instruction can be tailored to the individual, you’ll see learning progress increase dramatically.
N: You’ve said that top-performing companies benefit exponentially more from AI than companies that are less productive. Why? And what does it take to get a company to that level?
K: Well, you need resources to build the infrastructure, to collect and clean data correctly – and that’s one of the biggest obstacles. It’s boring, it’s tedious, it’s unfriendly. Nobody with ambition and talent and promise and says, “I want to clean data.”
But think about the importance of this investment. If you create a data model that’s going to move the needle just 1% for your business – that could to add up to millions of dollars by the end of the year. That’s profit that you wouldn’t have otherwise had.
If you create a data model that’s going to move the needle just 1% for your business – that could to add up to millions of dollars by the end of the year. That’s profit that you wouldn’t have otherwise had.
Here’s the catch: after you’ve tuned your model, it only takes a fraction of a second to run your analysis. You’ll spend 95% of your time getting the data ready before you can actually interact with it. It’s important to make sure that you have the foundation right before you start building.
N: What are the building blocks of a strong foundation?
K: First, you have to create a culture around data. Employees at every level of the company should be comfortable working with and listening to data. Find small questions that you can apply data to, and answer them as best you can. Failure becomes tolerable when you’re aiming for small wins. The pressure is off – and that’s what leads to big ideas. Build the mountain with pebbles as you agglomerate these small wins.
If you don’t have the flexibility to be creative with data, you’ll squelch innovation.
Next, you have to be flexible. Really wise managers will put aside a discretionary budget at the beginning of the year for spontaneous ideas that pop up throughout the year so that you can stay agile. If you don’t have the flexibility to be creative with data, you’ll squelch innovation.
N: That seems easy for the data scientists among us… but what can designers, writers, or other employees who may not directly work with data every day do to embrace it?
K: Understand that data can be used to help solve your problems – and learn how to speak the language of a data scientist so you know how to ask smart questions. Work on establishing that very basic level of comfort with data. You don’t have to be an expert, but you will need the confidence to interact with your data-loving peers, and determine how data can solve your problems when you harness it correctly.
Employees at every level of the company should be comfortable working with and listening to data.
N: Speaking of “smart” questions…What should we be asking right now about artificial intelligence?
K: If you’re a junior person in the organization, AI will be the future of your profession. If you’re not literate with machine learning and AI, you’re not going to be employable to companies in the future. Particularly as we start using algorithms to make decisions under conditions of uncertainty.
If you’re more senior, you should realize that you don’t need to be the “first mover.” You’re better off learning and being a fast follower. The talent behind AI is concentrated, and it takes a lot of time and imagination to apply new techniques to problems in ways that make sense. There are amazing research findings and early success stories of how companies like Google have succeeded in applying data – but it is still very uncommon. The world’s smartest engineers are earning millions of dollars to build the future of AI. This is not like playing tennis at your tennis club. This is like playing against Agassi or Federer – it’s an entirely different league.
So, learn what’s possible. Recognize that 90% of the people who pitch pie-in-the-sky solutions are just trying to feed at the trough of a red-hot industry. And once you see real examples of techniques that are working in the field, you can bring them into your own organization.
N: Thanks for chatting, Kenneth! Have a great rest of your data.
Take an even deeper dive into machine learning with our EVP of Data Science – then see how we use data to get s*** done at Red Ventures.
