The opportunity for machine learning in your enterprise
Machine learning is a genre of computing formed around computer programs able to learn for themselves without being explicitly programmed. It’s been around for decades, but only in the last few years has it become a solidly reliable and trusted vehicle to solve enterprise information management problems.
On May 11, 1997, an IBM computer called IBM ® Deep Blue ® beat the world chess champion after a six-game match: two wins for IBM, one for the champion and three draws. The success of Deep Blue charted the beginning of the machine learning revolution
The ML/AI hype-curve
You might be wondering ‘What’s the difference between artificial intelligence and machine learning?’ It’s more a question of ambition. Where machine learning is considered a software application that allows machines to learn from data without being programmed explicitly, AI is a bigger concept and it’s more about creating intelligent machines that simulate human thinking. The fact of the matter is, machine learning itself has still a long way to go in terms of its potential real-world value. We’ve only just scratched the surface. There’s so much more society and commerce can do with machine learning without moving to a TERMINATOR perspective of a future run by robots.
A recent study of more than 2000 work activities my McKinsey and Co. reported that about half of the activities (not jobs) carried out by workers could be automated using known technologies. They suggest, ‘…most workers—from welders to mortgage brokers to CEOs—will work alongside rapidly evolving machines. The nature of these occupations will likely change as a result.’
Spotting use cases
Check out our examples of where ML/AI is being used today to identify areas of opportunity in your business.
There are many ways machine learning can be applied in your enterprise to bring customers more value, improve customer experience, and reduce operating costs by automating data processing and speeding up time-to-value. That said, it’s not always obvious what SORT of tasks can be automated, so here are a few examples.
1. Seeing Patterns in Data
Marking and Checking
Machine learning is good at following a set of rules and learn from them. This has helped the technology to find a role in auditing documents in the legal profession, marking exam papers in education, qualifying topics and themes in RFPs so that previously used responses can be re-applied.
2. Horizon Scanning
Use ML/AI to spot opportunities and trends. Examples include:
Tender Response Management
Some B2B businesses generate the majority of their new business by responding to tenders—but responding to opportunities demands a great deal of human time and energy. Much of the work is about qualifying whether opportunities profile well with the abilities and credentials of the applying business. Machine Learning is well equipped to review Requests for Information (RFIs) and tender opportunities published by organizations. Profiles of ‘what an organization is good at’ can be etched into algorithms and then used to profile opportunities with similar attributes.
Customer Data Analysis
Image you want to understand patterns of behavior in the way customers interact with your business. This requires lots of information to be captured and examined. It may be the sheer volume of data that needs processing makes it uneconomic to have a human do it.
Market Opportunities
In the management consulting industry, providers offer guidance and expertise to clients as market conditions and regulatory frameworks change. Some of the larger providers now monitor events through news-feeds using ML to spot commonalities in the data and highlight opportunities for their services .
Tracking Brand Influence
In the media and Public Relations industry, providers offer ML-driven services to companies to monitor their brands in news articles; curating analysis to highlight positive and negative stories.
3. Monitoring Risk
Data Security
Sometimes, humans simply don’t have the time to process as much data as computers can; and they can’t perform the task 24/7 without a break. Machine Learning and AI is being used in data security provisioning where login behaviors of users can be vetted against learned behaviors that suggest a threat. Advanced systems can learn from experiences that resulted in a threat to reprioritize risk escalations.
IC Compliance
When employing new starters, it’s important that companies properly assess the employment status of candidates. GAO estimates U.S. federal and state agencies lose up to $57 billion due to workers being paid as contractors instead of employees. In the last 10 years, IC classification laws have been introduced in 30 states, including California and NYs “Freelancers Aren’t Free” Act, 2017. Today, ML/AI technologies for IC Compliance provided by companies like Greenlight, comb through contractor application submissions checking for back-taxes, interest, fines and penalties, stock options, 401K credentials and more.
Preventing Gender Bias
One challenge impacting on the lives and opportunities of individuals is the manner in which companies vet CVs from potential candidates. Historically, attitudes towards the employment of men and women have benefitted men. Filtering out gender bias phrases and terms in content to deliver a dispassionate and fair comparison between candidates is something AI is now doing.
4. Scenario Planning: Appreciating cause-and-effect consequences
Machine learning can be used to capture information on problems and issues, to then learn from those experiences to predict likely consequences.
Learning from Older Cases
This sort of use case can exist in case file systems where patterns of behavior need to be exposed. In areas of child protection, for example, it’s invaluable for case managers to be alerted to trends that have occurred in previous cases to highlight new threats.
Monitoring the Health and Uptime of Assets
Another occurrence exists in the asset management discipline, when the health and performance of assets needs to be closely monitored for potential downtime; or to predict capacity planning issues. For example, water companies have service-critical pumping stations located around the territories they serve. They now use ML/AI-driven approaches to implement preventative maintenance plans to ensure that their networks remain always online.
5. Optimizing or Displacing Humans
Perspectives on the impacts of AI often fall into two opinions on how the technology will revolve, often described as the IRONMAN or TERMINATOR end-games. With the first, AI technology evolves to equip humans to do better things. In the second, AI technology overtakes and subsumes humanity. As described earlier in this article, there are already thousands of use cases where ML/AI can assist humans to work in smarter ways.
Employees aren’t unsympathetic to the value of using ML/AI tools to help get their job done. More than three-fourths (76%) of the 1,000-plus survey group said having the digital tooling they need at work makes them more productive. More than half (53%) said it makes them more successful. The same share said they would be more empowered to better manage workflow if provided with the ML/AI tools they needed, and 42% said it speeds up boring tasks. Some 68% of the 18 to 34 age group said having the digital tools they need at work makes them more productive. That’s a significant share. But it’s even higher – at 80% – for the 35 to 54 age group. And a whopping 83% of workers age 55 and older agreed.
Source: Zenzar 2019 report
Medical Operations
Today, in surgeries around the world, AI is being used to ‘hold the scalpel’ for surgeries demanding a high level of movement control, sometimes accessing areas where the human hand could never reach (even if it were possible to hold your hand that still!).
Manufacturing
We’ve all seen images of robots on production lines taking over the roles of humans that once held a screwdriver and maybe even a hammer. Thanks to advances in hydraulic and pneumatic power, machines can outperform humans in the force they can exert, the consistency of their repetitive tasks, and they can learn and perfect patterns to a far greater extent.
Utilities Repair
Not all use cases are about displacing humans. In the majority of cases seen today, humans are AIDED not DISPLACED pay robots to work more efficiently. An award winning example of combining AI with drones exists in the utilities industry, where AI and robots are being used to repair fractured gas lines supervised by humans; addressing situations that would otherwise place humans at risk.
Planting Trees
Some people believe advanced technologies like drones and AI can be used for good—and here’s one example. The world needs trees, and lots of them. A UK company called Dendra is using machine learning to identify areas of the planet that offer an opportunity for planting trees. Then, they use drones and an innovative seedpod (containing a germinated seed and nutrients) to automate planting. Dendra estimates it can plant 10 up to billion trees each year – and at a much lower cost than the traditional method of planting by hand. The target is to plant 500 billion trees by 2060, in often hard-to-reach places.
6. Highlighting Irregularities
Spotting a needle in a haystack is not the most thrilling of tasks. One of the attributes of humans that is problematic sometimes, is our ability to get bored of repetition. The good news is that AI algorithms don’t worry about being bored and they can look out for patterns 24×7. Artificial Intelligence means that computers can learn from patterns in data and start to evolve ‘new learning.’
Betting Industry
One of the fastest growing areas of AI use is in gambling. Providers in the industry have to keep an eye on millions of lines of data every minute and spot any potentially fraudulent behaviors by using pattern recognition and AI.
Finding your Opportunity
In any enterprise, there will be use cases that offer early stage economies. If you can’t find great ways to use ML and AI technology, then you’re probably not looking hard enough.
To discover your ML/AI opportunity, consider:
- What mundane re-keying, spreadsheeting, data processing and ‘swivel-chair’ tasks are humans doing in your enterprise today?
- What data processing analysis, monitoring, and pattern recognition jobs don’t you have resources to perform at volume?
- What data processing needs to happen 24/7?
- What decisions in your business can be parameterized, to then be monitored by computers?
The ML/AI community began with a few hundred experts back in the 1970’s and ’80s’ but it’s not a ‘small and exclusive club anymore.’ The good news is that recent Open-Source projects like Rubix-ML are democratizing the technology, making it available to enthusiastic amateurs and professional data processing experts alike.
Is it good for society? That’s a good question, but one perhaps for another day.
