Automation or augmentation? AI in a company

In 2018, Unilever had already embarked on a conscious journey to balance automation and augmentation capabilities. In doing so, it has achieved impressive results – a 16% increase in the ethnic and gender diversity of new hires, savings of 70,000 working days per year, and a 90% reduction in recruitment time. But what are automation and augmentation? Let’s take a closer look, uncovering the dynamic interactions, opportunities and pitfalls, and the impact on business and individual employees. Read on to find out more.

Automation or augmentation – table of contents

What are automation and augmentation in the context of AI in a company?

Automation and augmentation are opposing but interdependent forces. In fact, companies face a choice: Do they cut costs and automate tasks, eliminating human involvement in the process? Or, with a focus on quality and personalization, enhance the capabilities of employees and improve outcomes through AI augmentation, which involves close collaboration between humans and artificial intelligence? Their complementary skills would then be combined to accomplish a specific task.

The paradox of automation and augmentation is an issue that modern organizations must confront. Understanding the difference and synergies between the two concepts is crucial for the successful implementation of AI in business.

Automation

Automation is the process of replacing human, repetitive activities with software. Before the era of the rapid development of generative artificial intelligence, automation was applicable only to routine and well-structured tasks, such as:

filling out invoices,
creating reports,
summarizing expenses,
simple customer service based on the selection of the next step of the conversation by pressing a button.

Organizations were able to automate processes based on expert knowledge encoded in the form of algorithms that define relationships between conditions (“if”) and consequences (“then”). Such automation was based on an explicitly defined domain model, i.e., a domain knowledge representation that optimizes a chosen utility function.

However, the development of generative artificial intelligence has brought radical changes to the field of automation. Not only can the new models respond much more flexibly to input data, but they can also execute commands expressed in natural language. In other words, instead of executing commands based on explicit rules, they can perform tasks based on contextual understanding.

Source: DALL·E 3, prompt: Marta M. Kania (https://www.linkedin.com/in/martamatyldakania/)

However, automations using artificial intelligence carry considerable risk.

The first is the dangers of automating decision-making – a problem faced by developers of autonomous vehicles, among others. For example, when a vehicle must make a maneuver in fractions of a second because there is no way to avoid a collision.

The second risk comes from relying on predictive algorithms. Even if a company would like to implement an automated option to follow data-driven artificial intelligence recommendations, a human must take responsibility for the decisions made.

A third type of risk is the use of generative artificial intelligence that, with insufficient data, begins to hallucinate, that is, to provide probable but false answers. For example, it may generate fake news or give customers false answers to questions. Navigating the benefits and risks of automation therefore requires careful analysis and preparation.

Augmentation

Augmentation is the process of using AI to enhance human intelligence and skills, rather than replacing them or acting independently. With the growing importance of augmentation in environments requiring complex decision-making, organizations are increasingly adopting this approach. For more complex tasks where rules and models are not fully known, augmentation enables natural and artificial intelligence to work closely together.

This is because augmentation is an iterative, coevolutionary process in which humans learn from AI and AI learns from humans. In doing so, the role of artificial intelligence should be designed to enable human oversight at all stages of a given process. It requires the involvement of domain experts, whose expertise is often tacit in nature, derived from years of experience and intuition, making it difficult or impossible for AI to directly replace them.

Augmentation allows humans and artificial intelligence to reinforce each other, combining machine rationality with human intuition, common sense and professional experience. This approach enables more comprehensive information processing and better decision-making.

At the perfume company, Symrise, for example, perfumers worked closely with the AI system to generate ideas for new fragrances (https://www.thefreelibrary.com/Can+AI+pass+the+smell+test%3F+Deploying+artificial+intelligence+can+be…-a0578441404). Through augmentation, experts were able to leverage the machine’s ability to process massive amounts of data while applying their own knowledge to interpret and contextualize the results. The results were innovative fragrances that customers loved.

Source: DALL·E 3, prompt: Marta M. Kania (https://www.linkedin.com/in/martamatyldakania/)

Smooth transitions – from automation to augmentation and back again

The relationship between automation and augmentation is dynamic. It allows for seamless transitions between the two approaches. The close collaboration between humans and AI within augmentation helps to identify rules and models that can then be used to automate a given task, leading to innovation and efficiency gains.

Organizations should therefore deliberately iterate between the separate tasks of automating and augmenting, making a long-term commitment to both.

Another step that will strengthen the link between automation and augmentation is the creation of autonomous agents, i.e. artificial intelligence that can not only automate tasks, but also plan processes and issue commands to other systems without human intervention. The development of next-generation AI solutions will also make it possible in the near future to create prototypes and innovative services based on needs analysis.

Summary

Automation and augmentation represent two opposing but often interdependent applications of artificial intelligence in management. A balanced approach that combines the strengths of both concepts is the key to achieving complementarity that benefits both business and society.

To manage this tension effectively, organizations should:

remember about the responsibility for creating transparent and secure systems using AI,
keep in mind the responsibility for management processes, treating AI as a tool to assist rather than replace managers,
integrate the two approaches by deliberately iterating between them and leveraging each other’s strengths,
implement strict controls and transparency mechanisms to detect and correct errors and biases in AI systems.

Above all, they should also invest in developing the skills and competencies of employees so that they can work effectively with artificial intelligence as part of augmentation.

Successfully combining these two AI forces will not only make organizations more efficient and innovative, but also help build a more just and sustainable society. The key is to understand that automation and augmentation should coexist in harmonious synergy, not compete as alternatives.

If you like our content, join our busy bees community on Facebook, Twitter, LinkedIn, Instagram, YouTube, Pinterest, TikTok.

Author: Robert Whitney
JavaScript expert and instructor who coaches IT departments. His main goal is to up-level team productivity by teaching others how to effectively cooperate while coding.
View all posts

Maintenance break

Automation or augmentation? Two approaches to AI in a company | AI in business #124

Automation or augmentation – table of contents