Feb 21, 2025 Andreas Fuhrich
Share"The engine might compose elaborate and scientific pieces of music of any degree of complexity or extent."
This quote from 1843 comes from the notes to a machine that was never built. Nevertheless, it is not insignificant.
Thanks to the blueprints of the machine, the Analytical Engine, we know today that it not only works, but would also have offered more possibilities than the Zuse Z1 developed almost a hundred years later (1937). However, the inventor of the Analytical Engine, Charles Babbage, lacked the capital and the necessary fine motor parts for his time.
The author of the quote mentioned at the beginning was Ada Lovelace. Elsewhere in her notes, she described how Bernoulli numbers could be calculated using the Analytical Engine. A description that is now regarded as the world's first computer code and makes Lovelace the first female programmer.
"The engine might compose ..." A generative machine, foreseen over 180 years ago, presented to the general public just over two years ago as ChatGPT.
Since then, we have been on the cusp of a new era of human-machine interaction. Applications such as ChatGPT are not only able to create content, they also understand the most natural form of human communication - our language.
Operating complex software with nested menus and hidden functions will be a breeze in the future. A simple natural voice command is all it takes to put an end to searching through confusing drop-down menus.
What are co-pilots and how do they differ from classic chatbots?
While classic chatbots are usually restricted to predefined scripts and limited use cases and quickly reach their limits when using language naturally, co-pilots can respond flexibly to a wide range of requests.
Transformer models and Large Language Models (LLM) form the technological basis. These models are trained on huge amounts of text and thus learn to capture the structure and meaning of language. Using techniques such as retrieval augmentation, co-pilots can access additional knowledge, for example from documents or databases, and integrate this into their answers.
They can also communicate with other applications and services via APIs to perform tasks. For example, a co-pilot in a CRM system can automatically send an email to a customer on request or enter an appointment in the calendar.
The benefits of co-pilots for software operation
The advantages of co-pilots in software operation are manifold:
- Increased user-friendliness: Interaction with software becomes more intuitive and natural. New users in particular find their way around more quickly as they do not have to learn complex menu structures. Voice control can also make it easier for people with physical disabilities to access software.
- Improved data analysis: Co-pilots enable quick and easy access to data and its visualization. Users can formulate complex queries in natural language and receive immediately understandable results.
- Context-related support: Co-pilots can understand the context of a task and proactively offer assistance.
Co-pilots: Possible applications
The integration of AI-supported co-pilots marks a significant turning point in the digital transformation of companies. The intelligent assistants will fundamentally improve the user experience with various applications and support employees in their daily work. They combine advanced technologies such as machine learning, natural language processing and data analysis to handle complex tasks and optimize decision-making processes.
Here are some scenarios from various fields of application:
- Combating fraud: Co-pilots could support analysts in investigating suspicious transactions. For example, an analyst could ask the co-pilot: "Which transactions in the last 24 hours show a similar pattern to suspicious transaction X?" The co-pilot would search the database, identify relevant transactions and present the results clearly.
- Logistics: In logistics, co-pilots could help to optimize supply chains, for example by planning routes on command, monitoring stock levels and predicting bottlenecks.
- Workforce management: In the area of personnel planning, co-pilots could make suggestions for replacement planning in the event of illness, taking working time rules into account.
- Aviation: In the aviation industry, a co-pilot could inform all relevant stakeholders of a delay based on the analysis of real-time data.
Challenges and limitations of co-pilots
Of course, there are some challenges to overcome before co-pilots take the business world by storm. One of the biggest is the tendency of LLMs to generate false or misleading information, so-called hallucinations, especially when the data basis is inadequate. Another important factor is performance: processing natural language and performing complex tasks requires considerable computing power. Ensuring a fast and responsive user experience despite large data volumes and complex models is a technical challenge that needs to be overcome.
Despite the many possibilities offered by co-pilots, even with their help computers will always remain just a tool to support decisions and increase efficiency, but they will never be able to replace human skills. Even if computers have the potential to compose something, there is still a limitation that Ada Lovelace recognized in Babbage's machine:
The Analytical Engine has no pretensions whatever to originate anything. It can do whatever we know how to order it to perform.
About our Expert
Andreas Fuhrich
Andreas Fuhrich is a freelance editor and co-author of various specialist books on digitalization. His main topics include the megatrends of our time such as the future of work, mobility of the future and artificial intelligence.