All Data is Not Created Equal – Four Ways to Avoid Bias in AI Algorithms

AI is taking on an important role in all areas of business – helping companies make critical decisions that affect people’s lives, such as who qualifies for a mortgage, how much an insurance plan will cost or how well a medical treatment will fare on certain populations. As I discussed in an article in Future of Sourcing, because of this responsibility, it’s more important than ever that biases aren’t inadvertently being built into AI algorithms.

What are some of the ways data professionals can avoid data bias when building an AI solution? Think of the following:

  • Hire a diverse team. The majority of data scientists today are white males; according to a Harnham’s Diversity Report for the U.S. Data & Analytics Industry, and only 18% of data scientists are women. Without diversity, there is a possibility for unintentional racial and gender bias to creep into the data.
  • Use diverse data. It’s important to constantly check the data for bias. Are mortgages being denied based on a person’s zip code, ethnic group or class? If so, make sure sufficient data is being added to provide a bigger picture.
  • Be transparent. Always be prepared to disclose the type of data used to train an algorithm and the criteria that helped make decisions, such as who gets a mortgage, or how risk is assessed.
  • Continuously test the algorithm. It’s critical to monitor your algorithm continually, not just to improve results, but also to make sure new data is not bringing new biases to the application.

AI is having a major impact on how decisions are being made, so data professionals must be continuously on the lookout for biased data that left unchecked can damage reputations, create legal implications and most importantly, decide the fate of individuals without sufficient data to support those decisions.

At Wovenware – We’re Open for Business

It has been nine weeks – and counting – since we last set foot in our corporate headquarters on Calle Los Angeles in San Juan, Puerto Rico. Like businesses everywhere, we had to quickly adapt to a remote way of working. And we’ve done that quite well. We continue to serve customers, while ensuring that the Wovenware culture remains intact. We’ve had a protocol for regular team meetings, employee check-ins for emotional well-being, and we continue to collaborate and actually have a bit of fun along the way. Check out the virtual musical performance – the Woven Music Jam COVID-19 edition.

Despite the silver linings, we all know video chats and phone calls can never replace real live interaction. Many employees have expressed a longing to get back to the office in order to have some sense of normalcy return. As of today, Wovenware is on track for a phased re-entry date of June 1.

As many businesses are doing, we’ve worked diligently to create a framework of standard operating procedures (SOPs) and protocols for that day and going forward to ensure the safety of our staff and visitors. So far, this is what we’ve planned:

  1. We’re establishing SOPs for all employees and visitors entering our corporate office, including required protective masks at all times inside the office. The SOPs and rules will be displayed on a poster outside the office.
  2. We will initially have a limited number of people in the office, starting with only up to 20 employees.
  3. We’ve purchased Wovenware-branded masks for each employee, as well as personal hand sanitizer and alcohol-based disinfectant for cleaning workstations throughout the day.
  4. We’re installing a hand disinfecting station next to the front door outside of the office.
  5. We also will be taking temperatures of employees every few hours.
  6. To maintain social distancing, we are moving workstations out so that they are all six feet apart from each other.
  7. We’re limiting use of common areas, such as the lunchroom, as well as all group gatherings until it is safe to do so.
  8. We’re reinforcing the importance of staying home if you are not feeling well or if someone you have come into contact with has tested positive.

While there will be clear-cut protocols for the office, our foremost concern is the well-being—both physically and emotionally—of our employees. We need to be flexible and understanding when it comes to employees with children at home and other family situations that are being impacted by the pandemic.

Like other companies in Puerto Rico, Wovenware will follow color-coded stages of return. Once the stay-at-home mandate (red zone) is lifted the company will move into the orange zone (workers will return who cannot do their jobs from home); followed by yellow zone (majority of workers) and finally the green zone when business is back to usual.

New information is coming out daily about the impact of COVID-19, best practices for combatting it, and dealing with the new normal, so flexibility will be key. What’s clear is that the return to the office will not happen overnight, but requires careful consideration and planning.

These are certainly strange times we are living in, and a new normal will emerge across businesses everywhere. What won’t be lost, however in this new normal is our commitment to our employees, customers, partners and the tech community. At Wovenware, we’ve always been open for business, but we’ll take new safe practices back to the physical space.

How the Woven Music Jam Project Started

As part of working remotely, it is essential to maintain constant contact with our customers, and more importantly, with our employees.

During the first six weeks of the pandemic, I had daily calls with all the developers whom I have assigned to me. In all those conversations, we talked about work, but also about life. Some of the questions that I asked were, “How is everything going?” “Do you have enough groceries?” “Are you doing exercises?” “Living alone, how are you managing this new scenario of life?”. I received all kinds of answers, and we supported each other.

Those first weeks were tough emotionally. Uncertainty took a toll on me, but at that point, I did not realize it. For reasons I can’t explain, I felt that one of the developers was lonely. His voice caught my attention.

I never asked him if something was wrong; I just decided that I wanted to make him feel happy. I wanted him to have something that inspires him.

Knowing that he is a musician and that he had not practiced in a long time, I decided to create a project where we could gather the musicians of the company to play a song together. The first one that I contacted was him, and he immediately said yes, and all the other musicians did too. Right at that moment, without knowing it, the Woven Music Jam journey started.

They picked the song in the first meeting, but the challenge was that most of them needed to practice before recording. They worked hard, and they wanted it to be perfect. We had eight planning meetings. They were always right on time to each meeting. In the beginning, it was weird for me. I thought that I did not have any skills to offer to this type project other than setting a schedules and pushing them to deadlines as a project manager does. I did not want to do that. I wanted them to enjoy the journey; to practice just for fun; to release stress; to and to feel happy.

At each meeting, new ideas were added to the project. I was responsible for hiring singers, so I even made a poster, but only a few, including me, accepted the challenge. At this point, the project was not only playing a song; it was an artistic expression. It was a gift from this talented musician to us and a gift to me.

Wovenware Jam Singers Wanted Poster

What I did not expect was the effect this project would have on me.

This project is an excellent reminder of how much joy I and others can feel in simple acts. We need so little to feel happiness but I believe that we need a lot of awareness to catch those moments and just be present in the moment. I was very present when I watched the first video. The feeling was overwhelming, and it happens every time I hear them play. They remind me how easy it is to work when you feel inspired. Most importantly, this project shows me how good the act of giving without expecting anything in return feels. They probably helped help me more than they will ever know. Thank you guys!

After you read this, watch the Woven Music Jam video. You may see it from another perspective. Hopefully you will feel joy in a simple way.

Big Data Analytics is Only One Part of the Decision-Making Process, But it’s the Key One

There has been much discussion lately about the role of data in making good decisions. Several industry pundits claim that while data is needed, human intuition, or gut feelings, ultimately have the final word in decision making. In fact, according to a Wall Street Journal article, a survey of CEOs from KPMG showed that just 35 percent of executives highly trust their organization’s data; and that two-thirds ignored insights provided by data analysis or computer models because it contradicted their intuition.

As I explained in a recent Forbes Technology Council column, effective decision-making doesn’t have to be an either-or-proposition – it just needs to lead with a hypothesis and then follow the data trail to reach the final decision. You can think of it as a data-journey.

Below are some of the ways good decision makers are blending their intuition with the data to reach better results:

Make sure the data can pass muster. As any good data scientist knows, once you form a hypothesis and bring in the data to support it, the data must be relevant and clean. Make sure you are determining if the data is it socially and financially valid.

Let the data change your plans. Once you’ve informed your hypothesis with strong data, be willing to change your plans based on where the data leads. Regardless of how sound you think your hypothesis is, if the data doesn’t back it up, you need to recalibrate.

Take baby steps. Even when you have a strong hypothesis supported by data-points that fully support the decision, change should not be made quickly, but rather, begin small, testing out the action over time so that you can continuously readjust and fine-tune accordingly.

While data gets better with time, so does intuition. Effective decision-making requires a big-data analytics approach to proving your intuition is right on target.

How We’re Working Through COVID-19: Best Practices to Keep Operations Humming and Employees Engaged

Today we are all facing challenging times personally and professionally as a result of COVID-19. When it comes to business, many of us are reinventing how we work, lead teams, communicate with customers and partners, and are finding new ways to gather around the water cooler with teammates.

Yet, for Wovenware, this isn’t the first time we’ve faced a major challenge. We learned a lot of lessons from our experience dealing with the aftermath of Hurricane Maria in 2017, especially about flexibility, agility and resilience. We recognized that the typical ways we were doing things wouldn’t work anymore, and we found new ways to adapt.

Ironically, while during Hurricane Maria we were working to keep generators working and employees in the office, today we’re working to keep them home. Yet, fortunately for Wovenware, our employees can easily work remotely given the nature of our business and our IT infrastructure which can support that.

Nevertheless, today we’re facing unchartered territory with the coronavirus, but we know that just as flexibility and resilience helped us get through Hurricane Maria, so too are we able persevere and achieve our goals now.

Our employees are the linchpin to our success – and I’m sure the same goes for all companies. During difficult times, it’s more important than ever to make sure that you are listening to the needs of your employees and supporting them any way you can. This goes a long way to creating a cohesive team that helps each other weather the storm and get through the hard times.

But, a big part of the challenge many companies are facing now is keeping operations running smoothly when remote working is no longer a corporate perk, but a business necessity.

Wovenware - Best Practices while Working from Home

Wovenware – Best Practices while Working from Home

Below are five best practices that have worked well for us:

  • Choose face-to-face communication whenever possible. Use video communication as much as possible – there’s nothing that is a substitute for seeing everyone’s faces. Video creates a feeling that you are all “in the same room,” enables more natural communication and boosts emotional well-being. In addition, because everyone can see facial expressions and body language, it can promote better understanding.
  • Over-communicate. While it’s important to make sure you are continually communicating with stakeholders at all times, when everyone is working remotely, it’s even more critical to overcommunicate with your team. The purpose of the calls is to check in with team members – personally and professionally – and keep the team apprised of activities underway. Likewise, it’s incumbent upon them to communicate status updates, especially when they finish a project or important task.
    If there are any issues that arise, handle them quickly through a video or phone call. When people rely on email, Teams Chat, text, IM, Slack, and other written methods of communication, it can lead to inadvertent misunderstandings. If you’re in doubt whether a colleague will want to know something, err on the side of sharing it.
    It’s also important to over-communicate with customers. They may be wondering what you are doing on their behalf and if you are able to be truly productive. Don’t wait for them to call you, but proactively call them often to let them know the progress that is being made on their projects, share written communications and best practices and demonstrate other ways you can be of service.
  • Be Proactive. To keep assignments on track and operations running smoothly, management should communicate clear expectations for team members. Establish daily and weekly task lists with the team, and make sure they are completed on schedule. At the end of the day, send out an email to the team listing all the activities that have moved forward. Check in to make sure team members have the tools and information they need, and if anything is impeding their work or their ability to meet their deadlines.
  • Protect Your Workspace. Speak with family members or other people in your household about the hours you are working from home and set up ground rules around the noise level, interruptions, etc. Even so, expect interruptions, such as a delivery person dropping off a package or your dog barking in the background. Try to anticipate these interruptions and avoid them as best you can, yet since we know we are all facing the same challenges, try to be understanding of any interruptions on their end.
  • Recognize the role of IT. The ability to communicate, collaborate and keep business running as usual largely depends on a well-functioning IT team. Include the IT team in general management meetings and let them know their significance to the company. It also gives managers an opportunity to share with them what is and isn’t working. When it comes to technology, be careful, not to deploy unnecessary changes to current systems, apply non-critical patches or install non-essential security rules. It’s not the time to introduce any potential glitches to the system.

While it is a challenging time right now, today’s telecommunications, mobile and other technologies enable companies to operate beyond the brick-and-mortar of a corporate office.  But success requires agility and paying attention to what matters most – your employees and customers.  We learned this from Hurricane Maria and we’re applying those lessons learned to the latest challenge – COVID-19.
We’re all in this together and together we will all get through this – and we’ll come out stronger and more resilient than before.

Getting on the Right Path for a Data Science / Machine Learning Career: The Courses and Sources to Get You There

Last year, I wrote a blog highlighting and suggesting a path to take for a career as a machine learning engineer. Some things have changed since then but many still remain the same. Data scientists, machine learning engineers or other AI-related careers still dominate the list of high paying jobs in demand. Unfortunately, what also remains is that the number of companies taking the AI plunge and integrating it into their processes is increasing while the number of qualified experts continues to be limited.

I couldn’t imagine a better time to become a data scientist. Data is the oil of the fourth industrial revolution. Everything from education to science is data driven.

Clients and colleagues continue to ask for guidance on how to start a career in machine learning or what skills are needed to augment and train existing personnel while AI projects are underway. Here is my updated review based on my experience growing our outstanding team of data scientists part of Wovenware’s AI Consultancy group.

The Required Course of Studies

So how do you get the advanced skills? There are many specializations in the AI and machine learning fields, and along with them come specific prerequisite studies, such as:

  • Analysis, Reading and Writing.  Data scientists must be story tellers. They must be able to insightfully read academic and industry papers.  They also must be good writers, understanding how to write for academia but most importantly, to communicate results effectively. This knowledge can be difficult to impart on someone, since it often comes through experience. This guide, however, can be a good start for understanding how to read academic papers.
  • Math and Statistics.  This includes courses, such as numerical analysis & forecasting, linear algebra, multivariate calculus, probability, regressions, and central limit theorem. Good resources include:
  • Programming.  The preferred programming language for machine learning and data science is Python, and some of the most popular libraries include, Pandas, Numpy, Matplotlib and Scikitlearn.  R continues to be the preferred language for statistics and exploratory data analysis, and some of the popular Comprehensive R Archive Network (CRAN) packages, are Caret, RandomForest and e1071.  Meanwhile, SQL is still a very relevant language, as relational databases are a big part of the enterprise. Data scientists must be fluent in these languages, while continuing to stay up-to-speed on new frameworks that continue to evolve.

Hitting the Books for Machine Learning Specialization

In order to gain expertise in specific segments of machine learning, it’s important to take specialized courses. A  popular and growing specialization right now is Deep Learning Specialization by Andrew Ng, one of the most influential and reputable experts in the field.  The course is offered in Coursera a learning platform that he co-founded.

A more general course in machine learning is Stanford’s ML class also by Dr. Ng. This one is more general but very insightful.

There are many others in Coursera. Another one is Udemy’s Intro. To Machine Learning lectured by no other than Sebastian Thrun and Katie Malone. This is a fantastic course in Udacity that offers nano degrees hosted by experts in academia and the private sector as well.

Machine learning offers many specialized areas and it’s important to select the field that best fits with your skills and interests. For instance, if you are interested in creating models for image processing, natural language processing and speech recognition, deep learning is probably the route to take. On the other hand, if you are looking to make predictive models to identify churn and customer tendencies, expertise in regression toolkits could be best for you.

I look forward to 2020, since it provides opportunities, as well as challenges related to deploying AI into production. These challenges will raise the need for new controls that enforce security and ethical policies, and this will require that even new skillsets be added to data scientists’ already bulging toolbelt. Stay tuned for an exciting year.