AI and machine learning are not a sci-fi pipe dream. They’re here, and they have been for a while.
However, up until recently, AI and machine learning lived in the background, baked into systems that few touched and fewer understood. Not only did it require enormous amounts of computing power (which no one had) but it also required vast fields of data to harvest and learn from.
But now AI and machine learning are here. The combination of enormous amounts of computing power available to literally everyone for almost pennies, and the vast amount of data that is generated, captured, and organized via smart phones has created a perfect storm for AI.
And this trend continues.
The Internet of Things, Smart Homes, Smart Cities, the Industrial Internet of Things, and edge computing are tributaries to a truly gargantuan data lake, then can then be mined to fuel machine learning, in term making AI outcomes a reality. In fact, that very data, and the AI that processes it, is arguably the competitive difference that sits at the heart of Amazon, Google, Facebook, and Netflix's runaway successes.
But that’s all old news. The rise of AI is not only well documented, but firmly understood at every level in businesses and organizations.
So the question naturally turns to how. Yes, Google, Amazon, and Facebook can turn that data lake into insights and, eventually, money. But how do other organizations, organizations not rooted in surveillance capitalism, capitalize on these innovations?
And, given that less than 25% of companies have incorporated AI into their product / service, there’s clearly still no simple answer.
To help prepare for the inevitable point where AI and machine learning are no longer a differentiator but rather, par for the course, here are our top 5 recommendations you should do to embrace AI and machine learning.
1. Organize your data
The #1 thing you can do from a technical level is organize your data. And that means understanding the state of your data now.
Data basically sits within a matrix that you can understand like this:
Unstructured data is data in your organization that machines can’t easily parse. Think PDFs or videos sitting on a hard drive somewhere.
Structured data is data that is organized and searchable by machine. Web pages showing up in Google search results are the best example of structured data. Machines can “read” (e.g. crawl) the page to understand what it’s about and surface the right content because it’s in a format designed for machines to do that.
Another, internal example of this searching for numbers in a spreadsheet. The numbers are formatted so that machines can search when you key in the figures you’re looking for.
Siloed is when different parts of your organization all access to different information, or worse, different copies of the same data. A good example of this is when an enterprise organization has multiple instances of a database, like CRM or a product data management tool (PDM) across multiple locations or divisions.
Aggregated data is when all your data feeds into a central repository and is pulled out as needed, which means that machines can pull data from anywhere across your org to use in algorithms.
For machine learning and AI to work, you want as much of your data to be aggregated and structured.
- Structured data gives AI and machine learning something to work with
- Aggregated data will give AI the potential to uncover insights that you can’t find manually
- Structured, aggregated data creates the volume of data needed to actually run machine learning.
Therefore, organizations should start working towards the data lake end point by investing in integrated solutions, data scientists and practitioners, and sounds data collection and governance.
2. Establish clear data protection processes and practices
Consumer, client, and member data is a precious commodity. More than that, personal data of any kind is a grave responsibility that organizations must manage with care. Not only do you need to manage data with care, but store, use, and leverage with care across their own and purchased tools / solutions.
The reason this is important is twofold.
First, consumers are less and less tolerant of privacy breaches. What’s more, the privacy breaches that do happen are having a larger and larger impact on brand value and consumer trust.
Second, the more data you aggregate, the more valuable it becomes for your organization… but the more valuable it is for others to target and steal. At the same time, the aggregation, structuring, and data exposure needed for AI and machine learning are the same forces that create vulnerabilities for your organization.
Fortunately, there are steps you can take now to get ahead of potential data problems.
- Build out a data protection team: Start building out your data protection and governance team, who will set policies, enforce decisions, and manage the structure of data, including establishing and enforcing best practices like limited access, data deletion, password management, and more.
- Make security part of your culture: The reality is, most security breaches are due to the humans at the heart of the operation. For example, a hacker can either try and crack a randomly generated, 99-digit password for months, or they can put on high-vis vest, stroll into your office, and just start doing work at a station, claiming “they’re from head office.’ That second option is a lot easier. Therefore, to prepare for AI, organizations need to think carefully about building a culture of security, where people hallenge things that don't seem right, treat security as a core part of their value add to the organization, and train regularly to understand and manage meeting threats.
3. Build an iterative culture
So far, we’ve talked about AI and machine learning like it’s going to just happen.
The reality, though, is that AI isn’t just going to happen.
It’s really, really hard. It takes a long time. And failure and rejection are a big and (often) necessary part of the job.
Therefore, you need to build a culture of iteration, trial and error, and of failure in order to leverage AI and machine learning.
And this isn’t just at the engineering level. This is at every level of your organization, from ICs all the way to executives.
4. Set expectations on your outcomes
If you’re a business, then you need to make sure that everyone understands what AI means for your organization.
And, just as pertinent, what it doesn’t.
For example, let’s say you make a software product, and you’re going to embed an AI engine into it to power a recommendation engine.
AI is going to make your product better, and (hopefully) you’ll make more money.
But here’s what AI won’t address:
- A product that doesn’t solve a real problem
- A go-to-market strategy that doesn’t have a targeted audience
- A product that doesn’t have product/market fit
AI and machine learning will make most products and services better. However, they won’t resolve underlying business or organizational problems.
5. AI / machine learning isn’t s silver bullet
This isn’t so much something to prepare for, but it’s something to keep in mind. Because AI and machine learning still have a little bit of sci-fi shine, it’s important to understand that neither of them is going to be a silver bullet for your product, service, organization, or business. Unless you are an AI developer who’s entire service is developing AI, then it’s only ever going to be a component of what you deliver.
What’s more, AI will increasingly be an unsexy piece of our product, solution, or service, more like the wiring in the car rather than the shiny rims that get you excited.
Ultimately, consumers, members, and businesses don't care about your AI or machine learning functions — they’re much too interested in what they get, when AI is really the how they get it.
Therefore, it’s important to treat AI and machine learning as what they are — extremely powerful and useful product features. Your business or organization will need to stand on its own, which means that you must resist the temptation to put all your eggs in the AI basket.
AI and machine learning are here. Not as a concept, or an idea, but here as a reality. Granted, it’s thin on the ground, especially once you move down from enterprise organizations.
That means there’s still a little time to understand AI and take advantage of it when you see an opportunity:
- Organize, aggregate, structure, and optimize your data.
- Establish strict data protection and governance policies for your organization.
- Build an iterative culture and a culture of failure from the bottom of the org to the very top.
- Set expectation for what AI can and can not do.
- AI isn’t a silver bullet.
In order to be successful, AI and machine learning are going to be part of the story — but not the whole thing. So don’t be scared of AI and machine learning. By moving now, you can still be an early adoption of the next wave of technology and innovation.