<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=956292202485176&amp;ev=PageView&amp;noscript=1">

How AI Can Help Humans Expand Their Skills in the Workplace

A woman looking at a tablet next to a large machine. The text reads, "Forbes."

This article, authored by DeepHow CEO Dr. Sam Zheng, was originally published on Forbes.


AI is rapidly outpacing Moore’s Law; its power and efficiency are on a recurring pace to double in fewer than 16 months, according to the researchers at OpenAI. This rapid progress is fueled by innovative algorithms (deep neural networks), big data (the internet and digitalization), and vast computing resources (availability of GPUs, TPUs, and AI SaaS).

The GPT-3 artificial neural network offers an interesting illustration of AI’s progression. GPT-3 is now capable of generating original and engaging short stories, songs, press releases, technical manuals, programming code, and more. It was released by OpenAI in June 2020 and has 175 billion parameters — 10 times the size of the former largest model (Turing NLG), which Microsoft introduced just four months earlier. By comparison, the human brain has roughly 86 billion neurons and has not grown appreciatively for tens of thousands of years.

As its capabilities expand, AI is transforming how we work. Indeed, excited by the enormous economic and financial potential of AI, thousands of companies are looking to adopt AI as a means to replace human workers. They should proceed cautiously because, unless they consider the needs and capabilities of humans, there are significant unintended consequences that can detrimentally affect overall business performance.

This phenomenon is often referred to as the "ironies of automation." A good example is Tesla’s manufacturing plant, where Elon Musk admitted that the company’s enthusiastic embrace of automation in its factories has actually put the brakes on production, tweeting in 2018: "Excessive automation at Tesla was a mistake. Humans are underrated." 

My background is in human factors and engineering psychology. I’ve studied the furious adoption of technology and factory automation as companies work to displace human workers from their jobs through digitalization and Industry 4.0. To date, most of these efforts have been focused on implementing technology that makes machines smarter; scant attention has been paid to making humans smarter through technology or how we can combine the best strength of humans and AI.

The pursuit of digital transformation has modernized manufacturing and introduced a wide array of advanced Industry 4.0 technologies that place new demands on workers and require them to acquire new technical skills in order to install, maintain, operate and optimize these new machines. There’s a lot to learn, and traditional training approaches are ill-equipped to keep up, resulting in a widening skills gap. 

On the factory floor, training techniques must assimilate large amounts of data that includes video, text, and diagrammatic information, and then codify workers’ complex manual interactions with complicated systems. Could an AI assist workers in acquiring this know-how? A number of manufacturers believe it can, and are dedicating a human-centric AI to the task of expanding the skills of both existing and incoming workers.

Supporting a company’s workforce readiness objectives while simultaneously positioning the workforce for the evolving future is no easy task. How do these manufacturers tackle the task? It starts with first assessing their learning and development priorities. These typically fall into four key categories: standard work and cross-training requirements, Industry 4.0 upskilling, knowledge transfer and conformity with Environmental Health and Safety (EHS), and quality mandates. (Of course, all training must factor in social distancing during the pandemic.) 

Next, for each of these, the training team selects the most appropriate training methodology. Not surprisingly, video is a top choice — especially with younger workers who grew up learning from YouTube how-to videos, not technical manuals. Realistically, then, this may entail creating hundreds of training videos — a time-consuming and extremely costly proposition because it takes, on average, seven hours to create a 1- to 5-minute video

Few businesses can afford an in-house video unit to comprehensively capture and document hundreds of tasks routinely performed by a skilled workforce. Here’s where a human-centered AI changes the game. It can observe and understand how humans perform expert procedures and tasks in a workplace setting, and then efficiently capture and digitize this expert know-how — distilling a skilled worker’s technical knowledge, know-how and dexterity into an effective "how-to" training video. 

In the real world, a purpose-driven AI dedicated to knowledge transfer can compress a half-day of video set-up and shoot into just 20 minutes. Once the editing process is complete, recipients across the globe can immediately watch that video with subtitles and voice-overs in their native tongue. This is knowledge transfer at scale and across language boundaries. It’s also AI helping humans grow into new roles that are more satisfying.

The timing for this is important to consider. In 2019, 50% of manufacturing workers were older than 44 years old, skewing heavily toward retirement age. In the coming years, as these millions of workers retire, they will take with them decades of valuable (some might say, priceless) expertise and know-how. We must capture and document this know-how today. At the same time, as new technology, driven by Industry 4.0 and advanced manufacturing, pervades across the shop floor, workers of all ages must be continuously reskilled or upskilled. 

Manufacturers recognize the urgent need to capture, document, index, and transfer this knowledge to the younger generation of workers who want a career where they can learn new skills and grow.



About DeepHow
DeepHow was founded in 2018 by a team of ex-Siemens researchers and engineers who saw an unmet need to transfer knowledge in the skilled trades labor market. They developed an AI-powered, video-centric knowledge capturing and learning platform that bridges the skills gap in the manufacturing, service, and construction industries. DeepHow streamlines know-how capture using AI workflow indexing and segmentation at one-tenth the time of traditional video-editing approaches, and powers knowledge transfer with smart, how-to training videos that boost employee performance by 25%. For more information, visit www.deephow.com/.

Recent Articles

DeepHow's Sam Zheng Joins Forbes Tech Council

DeepHow’s CEO, an expert in the application of AI to accelerate workforce readiness, will...

How to fix the $1 trillion skills shortage problem

This article, authored by DeepHow CEO Dr. Sam Zheng, was originally published on Forbes.


Uncovering the ROI of Employee Training

The landscape of employee skills and the requirements of the modern workforce are in a state of...

Sign Up

Get the very latest video-learning news delivered to your inbox.