What Is AI? Setting the Record Straight
Enterprise AI has reached a tipping point.
Whatever you call it – machine learning, predictive analytics, artificial intelligence, cognitive computing – this phenomenon is dominating our newsfeeds. It is interesting to see all of these terms being used interchangeably because subtle differences exist between them. For simplicity’s sake, we’re going to use them synonymously in this article, even though they’re not precisely synonymous.
It would probably be helpful to start with a baseline definition. AI is using a machine to understand past behavior in order to first predict, then potentially alter future behavior to produce more optimal outcomes.
All the major enterprise software vendors and hundreds of up-and-comers are rolling out AI-powered solutions.
Perhaps this business revolution is fueled by the ever-present AI we already experience in our personal lives: Netflix, Amazon, Pandora, Nest … even self-driving cars.
From before we even wake up in the morning, AI is making our personal lives easier through intelligent recommendations and automation.
Interestingly, the consumer space is 10 years ahead of the enterprise. Our employers have been slow to bring AI into the workplace, so there is currently a gap in the middle of our day where we largely exist without the help of AI.
Often, this means we are working harder and spending more time on tasks that could (and should) be automated or optimized.
Why has the enterprise been slow to adopt AI?
The reason it has taken so long for AI to permeate your workplace is that naive organizations believe AI is all about the algorithms and the math. But, in fact, the real value lies in the data and the workflow.
Meaningful AI requires three crucial elements:
1. Math – Algorithms that look for patterns in data in order to predict future outcomes.
2. Data – Information that continually feeds the math, making it smarter and more accurate.
3. Applications – Software that turns predictions and prescriptions from the math into improved outcomes by integrating into activities and workflows.
The math often gets the most attention. But, honestly, the math has been around for a long time. These days, algorithms are like opinions … everybody has one. The math has simply become table stakes.
The real magic of AI, and what sets winning solutions apart, lives in the data. Better data results in better predictions, which ultimately produce better outcomes.
Business-to-consumer (B2C) companies learned this a long time ago. Their advantage is larger customer bases that generate massive amounts of timely and accurate data. A great example is Amazon, which had already amassed 84 million unique shoppers as early as 2005.
Most business-to-business (B2B) companies struggle to generate enough scale in their own data to make it useful.
The only way to solve this problem is to passively crowdsource anonymous data across a wide variety of companies. In other words, we can only make AI work in the enterprise by working together.
To achieve the full power of AI, you also need the third layer: applications. Applications integrate the predictive and prescriptive recommendations from the AI platform into the workflow of employees.
Without the right applications, you get what we call “No-lift AI.” It may be predictive, but it doesn’t drive real business results because employees don’t use it.
Workers ignore AI for three main reasons:
1. They don’t trust the recommendations.
2. They think they know better.
3. They’re lazy.
When you bring all of these elements together – math, actionable data, and apps – you can achieve substantial business impact.
Dave Elkington is the CEO and founder of InsideSales.com, the industry’s leading cloud-based sales acceleration technology company. All of InsideSales’ innovative products are fueled by the predictive insights of Neuralytics, a self-learning engine with more than 100 billion sales interactions and counting.
Revenue Growth @ myBillBook | Leading high performance Marketing & Inside Sales teams | PLG & PLS
8 年Most enterprise problems can be solved by a simple classification algorithm. You would be surprised how the good old Random Forest works against GBM and other blackbox ML. The problem is data integrity since a lot of data is still manually entered, tweaked or isn't true because of some other reason. For example you wanted to understand which email template gets the best connects. Maybe a critical signal in this scenario is the designation of the person receiving the person. Unfortunately a Sr. Manager in one company is a decision maker as opposed to a VP in another. Imagine that this kinda random fluctuation happens very usually across companies and in other categories of your data. For example, you also try to use company category as another signal. This data comes from another source like Mattermark or Clearbit who have their own algos to classify companies. So a holiday packages site is a deal business and a hotel site is in hospitality. Some businesses are missing a classification because their site description is missing. So the algorithm tries to make sense and find combinations which give you the least false positives. But sadly garbage-in-garbage-out. Meaningful AI requires one critical input. Super accurate data. The rest is as simple as uploading a file in BigML or Azure :)
Husband I Dad I Entrepreneur I Partnerships I Coach
8 年And with OPTYX from ChannelEyes, you can now put the power analytics right in the hands of your channel sales team, not just your direct sales team.
Guiding customers in thinking about new ways to engage with buyers, grow revenue and institutionalize modern sales solutions.
8 年Interesting thought leadership on AI by the team at InsideSales.com, they have helped us transition to consuming the most accurate data versus being consumed by data. An influx of streaming data can be paralyzing, AI can interpret the story versus reading the story back to you.
Jeff Hawkins defines Intelligence as the ability to predict the future
Executive Sales Consultant
8 年You need all three: math, data and apps to realize the value from AI, like three points on a triangle-spot on!