The Downside of AI Powered Transformation — Part 2

By Seth Earley  with Richard Lynch

AI DispacementIn Part 1 of AI powered transformation, we illustrated the upside of  Artificial Intelligence (AI) in healthcare, safety, robotic process automation, and other knowledge intensive work such as  behavioral health cognitive agents or “bots”. Benefits to society such as improved health outcomes, increased access to therapy and faster response time to emergencies will also have significant costs — the primary one being that of job displacement. Forrester estimates that by 2025, 12 — 18% of jobs will be lost and those displaced will not have the skills needed in the advanced digital economy.  

As jobs are automated and workers are replaced, the remaining jobs will likely experience wage suppression. This trend will further the economic divide and potentially lead to a less stable society. 

Since computer algorithms are behind much of what is called artificial intelligence, it will be important to understand how those algorithms function because they are based on assumptions and judgments made by their programmers. Therefore, there can be hidden biases. The same problem can be attacked by algorithms with different core assumptions and lead to different outcomes. They are not objective and it will be increasingly difficult to identify biases and the impact of assumptions as more AI’s work together or process inputs from other AI’s. This increase in complexity will make it more difficult to tease out or understand the impact of interactions and hidden biases.

One implication for federal and state government is that they will need to use unbiased intermediaries and objective AI benchmarking approaches to understand AI capabilities, evaluate approaches and develop policies related to AI. 

On the economic front, there will be threats to the current structure and opportunities to improve economic outcomes. As simpler processes are automated and complex repetitive service requests are taken over by bots, jobs will inevitably be lost. At the same time, the same tools that will take over repetitive tasks will be used to advance skills and train workers to take on more complex tasks. Customized skill development programs will be able to adapt to an individual’s learning approach and fill in missing knowledge as they build advanced knowledge and capabilities. 

Creativity and problem solving will remain within the realm of human capabilities for the near term. The key will be to give people the means to pursue passions with more free time that will become available.

Trade jobs will remain in high demand since these skills will resist automation for the foreseeable future. One of the challenges regarding displacement lies in the wide gap between what the education system provides and the emerging demands of industry. Though traditional library science is in decline, library science integrated with knowledge engineering will be in very high demand. Universities are ill-equipped to adapt to the rapidly changing skill demands of industry. This means that industry will need to develop their own knowledge engineers and build knowledge engineering approaches optimized for various disciplines and domains. 

Due to the nature and scale of disruption, the AI Revolution will be messy and painful — even more so than the industrial revolution.

Implications for Policy Development

Corporate and Government policy needs to be developed proactively in several areas. When investigating or adopting tools and technologies, it will be important to have a mechanism for benchmarking functionality and detecting biases. Standard data sets and defined use cases can be developed for benchmarking purposes. Algorithms will require some level of auditing to understand the impact on a process. Many such algorithms are already in use – such as predictive approaches used in criminal justice. As other applications become more widespread, a process for understanding the impact of outcomes will need to be in place.

On the regulatory side, state government will want to encourage innovation while monitoring public safety. The development of driverless cars is one example, adoption of tools to improve efficiencies within state government need to be encouraged, not limited through regulation. 

Data, content and knowledge processes need to be properly managed to develop AI capabilities. This means that the corporations and government need to get its own data house in order. Along these lines, organizations need to be innovative in the whey they retrain workers to prepare for new jobs. The entire system for delivering services can be transformed through digital innovation and made smarter, faster, more effective and more efficient. Replacing routine, repetitive jobs with machines  is just the tip of the iceberg. AI is replacing customer services, coaching, training, behavioral health case management, and other jobs. Workers need to learn to partner with AI agents and bots.

The acceleration of job transformation and worker skill development needs to be increased with attention to and understanding of career paths during disruption. Rather than training people to execute a task, transformation efforts must include training how to think and solve problems.

For more information on Government's role in AI

[Read this blog]

About contributing author Seth EarleySeth Earley is Editor, Data Analytics, IT Professional Magazine from the IEEE and CEO of Earley Information Science (EIS). His interests include Knowledge Strategy, Data and Information Architecture, Search-based Applications and Information Findability solutions. Seth has worked with a diverse roster of Fortune 1000 companies helping them to achieve higher levels of operating performance by making information more findable, usable and valuable through integrated enterprise architectures supporting analytics, e-commerce and customer experience applications.

 

Share: