Will Deep Learning Technology Disrupt Management?
Are you prepared to work with AI-powered ‘prediction machines’ to run your division?
Managers diagnose problems and opportunities, classify them into patterns, and develop action plans. Diagnosis is the most important first step as a leader and executive. Recent advances in Deep Learning are revolutionizing diagnosis and classification in many fields.
Last week, I attended the annual NVIDIA GPU Technology Conference in Washington, DC. There was an amazing display of technologies powering machine-aided prediction (diagnosis) and autonomous systems. Presentations covered advances from healthcare diagnostics to satellite imagery analysis and many fields in-between.
One of the hottest topics was Deep Learning applications. Deep Learning is a type of Machine Learning within Artificial Intelligence (AI). Machine Learning techniques “train” a computer model to discover patterns using labeled input data. The software adjusts its internal values until it finds the best prediction model. The training examples may range from hundreds to hundreds of millions. This approach contrasts with traditional programming. In traditional software development, programmers develop heuristics and algorithms to interpret the data. They then write explicit, rigid instructions for the computer to follow (e.g., ‘if/then’ rules).
There Was One Field Missing at the Conference — Management
Can a technology that is disrupting cancer diagnosis revolutionize the field of management? The NVIDIA conference had a track on Deep Learning applications in medicine. Researchers presented advances in Deep Learning for Radiology and Pathology diagnostics. They showed results where machine predictions were near the best human diagnosis. Deep Learning presentations are crowding out other topics at large international medical conferences. Isn’t the field of management also dependent on diagnosis and prediction too?
Can Deep Learning make a good manager a great manager? Successful general managers (GMs) and program managers are effective in making predictions. They use their experience and reasoning ability to express judgment. They foresee changing forces, forecast results, and adjust. They enable their team to see customer needs before the competition. GMs depend on diagnosis and forecasting by specialists in finance, sales, marketing, human resources, supply chain management, and operations. They encourage creativity and continuous improvement to increase productivity and profits. Great GMs see emerging patterns, predict the implications, and lead teams to act.
- What if you could improve your management diagnosis and prediction abilities with Deep Learning?
- What if it became inexpensive to make sophisticated predictions about sales opportunities, customer behavior, staffing decisions, and process improvements?
- Will customers need your firm’s experts if they can get it from a Deep Learning-powered prediction machine?
- What if this technology was available to your competitors? Your customers?
Seeds of Enthusiasm for Deep Learning’s Potential
I was a systems engineer and programmer before becoming a program manager and General Manager. I started my career building computer models to simulate real-world systems. I had the opportunity to experiment with the precursor to Deep Learning in 1987 at TRW(now part of Northrop Grumman) — they were called ‘artificial neural networks’ then. The technology held enormous promise but was immature. I expected rapid advances and adoption. I was wrong about how long it would take to perfect the technology and techniques. I have been intrigued ever since.
Fast Forward Thirty Years…
Deep Learning is now ready for primetime applications. Today’s Deep Learning renaissance is now possible due to data abundance, cloud computing, GPUs (graphics processing units), algorithm advances, and open source software.
I’ve spent the last three years diving deep into the field of data science and machine learning. I’ve taken courses, helped a university set up an analytics program, and built computer models. I have experimented with the latest open source tools and technology. The field and applications are growing fast.
I’m amazed by the ease of use and pace of improvement in tools. There is a 24x7x365, dynamic, and global community of people sharing their experiences and how-to advice. It is open to everyone with a computer and internet access. You no longer have to work at a large company or research institution to have access to this emerging technology.
Will Deep Learning Power our Economy’s Lagging Productivity Growth?
We are on the precipice of significant improvements in productivity. It will take place across a wide range of knowledge work. It will disrupt and complement the work of white-collar workers. Deep Learning will power it. The benefits to society will be enormous. Expect exponential growth like other digital technologies.
Revolutionary changes in many expert professions will emerge. Prediction technology will complement human judgment and creativity. Economists predict shifts in what we value as the cost to make an expert diagnosis drops. Entrepreneurs will develop new categories of products and services powered by Deep Learning.
NVIDIA GPUs and cloud computing will make the processing power abundant. The Internet of Things (IoT) will make the data needed to train the models abundant. Open source software will enable creative minds across the globe to find new uses. Global technical collaboration platforms (e.g., stackoverflow.com, github.com) and MOOCs (e.g., Coursera, EdX, Udemy, Udacity, etc.) will grow the talent pool and speed advances. New startups will simplify the tools so business analysts can use them.
Value and Wealth Will Emerge for those that Address the Bottlenecks
NVIDIA and cloud computing companies are addressing the computation and storage bottlenecks. Universities, MOOCs, and internet-based technology knowledge sharing sites are addressing the human capital bottlenecks. Niche technology firms are adding value to open source software with technical support and specialized add-on features.
Two major bottlenecks remain. Addressing these will create more wealth and growth opportunities. They are training data and imagination.
Deep Learning models need large volumes of high quality curated data for training. Those who control this valuable data will generate enormous wealth digitizing, curating, and licensing their ‘labeled’ data. That is already happening and will grow. Many organizations are only now realizing the value of their information stores. For many, the digitization and curation work will be labor and time intensive but worth the investment if imaginative managers and entrepreneurs can discover new insights and applications.
Finally, the toughest bottleneck will be one of creativity and imagination. Managers will need to re-examine their processes and look for ‘expertise bottlenecks.’ They will need to sponsor experiments to test Deep Learning applications. Developing and attracting talentthat understands this technology will be critical. Imaginative entrepreneurs will create new services enabled by the disruptive economics of Deep Learning. The future is very bright for those that have the imagination and leadership ability to experiment with this revolutionary technology.
Intrigued by the Opportunities for Deep Learning Applications?
Primehook Technology helps our clients innovate by testing emerging technology and new management ideas. We work with general managers to upgrade their strategies, structure, and talent. We help them take these innovations to market. Let us help you explore prediction technology applications, harness data abundance, train your staff, and develop an innovation-powered growth strategy and structure.
_____
This was originally published 11/10/2017 here: https://www.linkedin.com/pulse/deep-learning-technology-disrupt-management-leif-ulstrup/
