Pioneering AI Frontier: Dynamically Reconfigured Business

In the first two articles of this series on the future of Artificial Intelligence (AI), I explored the possibilities and challenges of using applications with natural language interfaces and access to integrated knowledge bases. In this third and final article of the trilogy, I will discuss how AI could evolve to dynamically reconfigure business processes and policies in pursuit of better outcomes by continually redesigning itself.

*DRAWING HANDS (DESENHANDO MÃOS) – MAURITS CORNELIS ESCHER*

Note: This series does not seek to predict the future, but to design it from what we want, thereby guiding technological evolution initiatives in organizations.

Workflow Management

Many organizations today operate their business processes using workflow platforms, where automatic and human tasks and activities are controlled by computer systems to flow according to a sequence planned by experts to deliver the expected outputs in compliance with the required quality controls.

This is how practically all processes of large companies occur today, from hiring employees, paying salaries and suppliers, issuing invoices, customer service, purchasing materials… everything that happens repetitively and that is structured and regulated in some way.

Automated workflow - Created using the AI Midjourney — *Automated workflow – Created using the AI Midjourney*

A change in such a flow depends on an organizational transformation project, where business analysts evaluate the gaps in the current process based on performance indicators, identify possible root causes and opportunities for improvement, redesign the process with the necessary changes, and forward these changes to systems analysts and developers who version models and codes to modify flows and rules that will be tested before being updated in the production environment. In other words, each change involves a considerable effort by several people to modify the work routine.

Imagine a company where people, robots, and AI systems continuously collaborate to adjust operations in real-time, as part of the routine. Not as a project, but as a process of continuous improvement.

In the future, in addition to being used as an automatic task execution resource, AI will also monitor and redesign processes and rules dynamically in pursuit of implementing strategies and achieving goals defined by humans. AI will not be limited to rigid policies and pre-defined activity flows, but will be able to adjust automatically to optimize outcomes by modifying policies and flows.

An artificial intelligence that programs itself in an infinite loop, created using AI ChatGPT-4 and DALL-E, Open AI

An Example from an Airline

Let’s use as an example a process of an airline that happens daily. When a flight is delayed, several passengers miss their connections on other flights that were previously planned. From this event, a set of activities needs to be performed:

Identify all passengers who have connections from that flight;
Check the schedules of each of the connecting flights to identify which can still be maintained and which will be lost;
Reallocate passengers from lost flights to the next flights at the most convenient time;
Redirect their baggage to the new connections;
Cancel their previous tickets and generate new tickets;
Properly guide the affected passengers about the change by the crew members.
In case of long waits, they need to be allocated in hotels and receive meals or even special attention due to health conditions.

In this process, a set of rules and policies define criteria for operational decision-making. For example:

What is the minimum time to enable a connection?
How much does the passenger receive in meal vouchers?
What is the minimum waiting period that requires the passenger to receive free accommodation in a hotel room?

However, many passengers come up with conditions and preferences not foreseen in the process, generating moments of stress alongside powerless employees at the company’s service desks. (I hope you have never been through this, but those who have know what I’m talking about).

What to do if the passenger is a wheelchair user?
And if it’s an unaccompanied minor?
An elderly person?
What if there is no available space on the next flights?

Each of these situations requires analysis and development of new tasks and rules that will be the focus of an automation project so that they can be treated in a structured manner when they occur again. It is through multiple projects that workflow systems evolve.

Dynamic System for an Airline Using AI

In the future, these and many other processes and related rules will be continuously monitored and adapted by artificial intelligence. Instead of analyzing, redesigning, and automating the process, the human work as a manager will be to define strategic objectives to be continuously optimized by AI. Example:

Maximum passenger satisfaction
Minimum operation cost

In the case of possibly conflicting objectives like these two above, the manager should clearly indicate the priority and possibly weights for finding the desired balance. AI will take on the responsibility of redesigning the processes and configuring the rules in search of these outcomes dynamically and continuously.

In the future, the difference between airlines may no longer be how well they implement their strategies since they all will possibly use a similar AI to be excellent in their implementation. The difference will be the weight they put on each of these objectives, which will be directly reflected in the prices of tickets in search of different customer profiles.

In theory, this should already be happening today, and the higher price would mean a superior level of service. But we know that between strategy and execution, there is a true abyss of poorly modeled processes and rules and unforeseen situations that generate anger and frustration. AI can help considerably reduce this distance with extremely optimized services.

An example of a Smart City

AI also has the potential to revolutionize the formulation of public policies. Governments can use AI to analyze data more comprehensively and predictively, identifying emerging social, environmental, or infrastructure problems, setting goals, and implementing more effective policies. In a smart city, sensors for traffic, people’s transit, and pollution will continuously measure indices and, based on the information collected, promote various improvements such as:

Optimize traffic lights and signals to reduce traffic jams, improve flow, and minimize time spent at traffic lights.
Predict congestion and offer alternative routes to drivers, based on events such as accidents, roadworks, or weather conditions.
Provide information to drivers about where to find parking spaces.
Adjust parking rates based on demand, encouraging the use of public transportation or carpooling to reduce traffic and pollution.
Optimize public transportation routes and schedules based on real-time passenger needs. This includes adjusting the frequency of buses, trains, or subways and providing accurate information about arrivals and departures.
Identify and fine vehicles that emit pollutants above permitted limits.
Restrict traffic in high-pollution areas during air pollution peaks.
Adjust energy distribution to optimize the use of clean energy sources.

This list is not even about something futuristic. These are real examples of what several smart cities are already doing today with the use of AI at different levels.

Many other benefits can be implemented in the future in areas such as public safety, epidemic prevention, citizen service, cleaning and maintenance of public areas, tax collection, education, and wherever there are processes and rules.

Challenges for the Development of Dynamic Systems with Artificial Intelligence

However, this AI revolution is not coming without challenges and concerns. See below 4 aspects where the implementation of dynamic systems with AI still needs to evolve and why.

Technical

The language models available on the market today have proven to be very skilled at writing source code for software applications or describing business processes and rules in a structured way, but they are far from infallible. On the contrary, errors are still more common than correct codes, and generally, the professional use of this information has required huge human intervention. Even so, it is undeniable that they are being used as tools and incredibly optimizing the productivity of those who use them.

It is difficult to predict if at some point we will have totally reliable systems to monitor and configure business processes and rules autonomously in the way I propose in this article, but we can certainly count on AI systems in a “supervised” form for this to happen. In this form, systems present the result of their analyses and the change recommendations and wait for authorization from the responsible human before they can be implemented. If necessary, this human could make manual adjustments.

Just as it happens with a new employee, trust in AI must be gained over time, and the level of autonomy delegated can be reevaluated incrementally.

Ethics

As I have already commented in the previous articles, AI has the potential to amplify prejudices and biases present in the data used to train it. Especially if the decisions made by AI feed back into the training process, which exponentially potentiates tendencies.

Ensuring that AI-based systems make ethical and fair decisions is an ongoing challenge and requires monitoring and the creation of “guardrails” (protection limitations) that prevent AI from crossing certain boundaries.

Regulations and legislation that establish these boundaries in international agreements are already becoming a reality, such as the Artificial Intelligence Act of the European Union, but there is still much to advance.

Privacy

Another point that deserves attention is how to deal with generalized access by sensors, cameras, and shared databases when dealing with sensitive data.

Privacy issues need to be discussed, and some guarantee must be given that this information will be used by the authorities only in the legitimate interest of the citizen and not offensively to him.

In addition, it is necessary to ensure that this information is stored and maintained in a secure and restricted manner. The security of AI systems is essential to protect confidential information.

Equity

Access to and adoption of technology is not uniform. While we are here discussing the creation of smart cities and dynamic systems, according to UNESCO, 46% of the global population still lives without access to basic sanitation. Economic and social inequality is already abyssal in our world.

With the increasingly rapid advancement of management processes powered by AI tools, the survival of organizations and countries that fall behind may become unviable, unable to face inequalities and compete in the market. This tends to generate even more centralization of economic and political power in a few organizations.

Policies and regulations that require sustainable development with a focus on the entire planet, such as the Sustainable Development Goals proposed by the UN, need to be supported and considered on the list of objectives of all projects involving AI. It is not about stopping the evolution of technology, but ensuring that this evolution occurs sustainably and for social and environmental benefit.

Conclusions

Although the future of AI brings a set of challenges, I consider it to be promising. AI is redefining the way we operate our businesses with the potential to improve our lives in numerous areas, and we need to engage with this development to ensure that our interests are prioritized.

In this series of 3 articles, I explored possible advances that we might achieve with the use of Artificial Intelligence systems:

Modifying the way people interact with machines;
Reorganizing and maintaining our knowledge bases alive and up-to-date;
Dynamically optimizing our organizations.

Many of these evolutions will need our support and participation. We should not just be spectators of the changes in the world around us, but protagonists in our sphere of influence.

I invite you to be part of the transformation and to discuss these challenges in your organization, community, or the projects you are involved in.

What do you expect for a future supported by AI solutions?
How would you like AI to impact your day-to-day life?
How can AI be used to make the world a better place?

Share your thoughts and participate in the construction of this exciting future. Thank you for following this series of articles about the frontier of AI. Let’s shape tomorrow together!