Artificial intelligence is one of the rising trends in all areas of discussion, online or otherwise. That is why making your way within the “noise” of the constant flow of information from more or less reliable sources is increasingly complicated. It is worth exploring the relationship between Artificial Intelligence and Business Analysis, but gaining clarity is the first step to better understand what we are talking about.
In the beginning was the Algorithm
An algorithm is a set of detailed and ordered instructions for performing a specific task or solving a problem. One can think of an algorithm as a cooking recipe: a clear and defined sequence of steps that, if followed correctly, leads to a specific result or product. In programming, algorithms are the fundamentals of how software and computer systems work.
An algorithm begins with an input, consisting of data or initial conditions. Through a series of steps, an output, which is the solution to the problem or the result of the task, is produced. These steps are generally sequential and must be performed in a specific order to achieve the desired result.
Algorithms are ubiquitous in modern technology. Their effectiveness is determined by the accuracy, efficiency and speed with which they can process the input and produce the output. A good algorithm is able to perform its task as efficiently as possible, reducing processing time and resource use.
Despite the leaps and bounds of technology in recent years, the concept of an algorithm belongs to the 19th century.
Forerunner of the times: Ada Lovelace
Ada Lovelace (London 1815 – 1852) is universally recognized as one of the first figures in the history of computing. The daughter of the famous poet Lord Byron, Ada was educated in mathematics and science, an unusual path for a woman at the time, thanks to the encouragement of her mother.
Ada Lovelace’s most significant contribution to the field of computer science came through her work with Charles Babbage, a mathematician and inventor who was developing the “Analytical Machine,” a mechanical device that anticipated modern computers. Lovelace was fascinated by this project and began working with Babbage in 1840.
Her most famous contribution was the publication in 1843 of a series of notes on the article by Italian mathematician Luigi Federico Menabrea. The text concerned the Analytical Machine: in these notes, Lovelace not only translated the article but expanded it, adding his own comments and ideas. The most notable of these notes, labeled “Note G,” contains what is considered the first algorithm intended to be processed by a machine. In this note, Lovelace described a method for calculating Bernoulli numbers with the Analytical Machine.
What distinguishes Lovelace is not only the fact that he wrote an algorithm, but also his view of the potential of the Analytical Machine. While Babbage himself saw his invention primarily as a calculating machine, Lovelace sensed that it could go further. In his vision, the machine could process not only numerical quantities but also symbols and potentially creating music, art, or other kinds of nonnumerical data. This vision anticipated what would become the modern concept of programmable computers and software algorithms.
How did we get to Artificial Intelligence?
The main distinction between simple algorithms and Artificial Intelligence (AI) algorithms is based on their complexity, autonomy, and learning ability.
Simple algorithms are step-by-step procedures or formulas designed to perform specific tasks. They are deterministic, meaning they follow a predetermined set of instructions and produce predictable results. They are static, do not adapt or learn from their previous executions, and are limited to tasks for which they were specifically programmed.
AI algorithms, on the other hand, are designed to mimic human decision making and learn and adapt based on the data they process. AI algorithms such as those used in machine learning can handle complex tasks that require nonlinear processing. They have the ability to learn from data, improving and adapting their responses over time. This makes them ideal for applications requiring advanced analysis, prediction or autonomous decision-making. Their potential is manifested in their ability to learn directly from data. With exposure to new information, these algorithms are able to refine and improve their performance over time. It is a continuous process of learning and adaptation that makes them effective in applications of advanced analytics, accurate predictions, or autonomous decisions.
Machine Learning
Machine learning is a field of AI that focuses on the ability of systems to learn and improve autonomously from experience without being explicitly programmed. Using algorithms that analyze and interpret data, AI applies patterns and behaviors by making predictions or decisions based on that data and projections of it, being able to apply calculations and processes even to unexpected or unanticipated contexts.
There are three main types of learning in machine learning.
Supervised Learning
In this type of learning, the system is trained on a pre-labeled data set. This means that each input data in the training set is associated with a correct answer (or label). The system’s task is to learn a model that can make accurate predictions or classify the data into categories based on these examples. For example, in an image recognition system, photographs are labeled with the name of the object represented, allowing the system to learn how to identify objects in new images.
Unsupervised Learning
Unlike supervised learning, unsupervised learning works with unlabeled data. Here, the system autonomously seeks to discover hidden structures, patterns or relationships in the data. For example, the system might group data into clusters based on non-obvious similarities. This type of learning is useful for discovering non-obvious information in the data, such as market segmentation in marketing or anomaly detection.
Learning by Reinforcement
This form of learning is based on a process of trial and error, in which the system learns to perform certain actions to maximize some sort of reward or minimize a punishment. The system receives continuous feedback that guides its decisions. For example, in a game of chess, the system learns which move leads to the best outcome (checkmate) through a process of rewards (winning a game) and punishments (losing a game).
Generative AIs
The distinction between “normal” AI and generative AI is based on their functions and output modes. “Normal” AI includes systems designed to analyze, interpret and make decisions based on data. These AI systems are commonly used in applications such as natural language processing, where they interpret and respond to human language, in computer vision for object recognition in images and videos, and in predictive analytics to make predictions from historical data. The main goal of these systems is to process and analyze existing data to perform specific tasks such as classification or prediction, using supervised or unsupervised learning models.
Generative AIs focus on creating new data or content that did not exist before. These advanced AI systems are capable of generating text, images, sounds and other types of content that can emulate human creations. For example, Generative Adversarial Networks (GANs) use an approach in which two neural networks, one generating and one discriminating, work in competition to create new and compelling content (e.g., creating and destroying password attempts to calculate their complexity).
Generative text models, such as GPT, are also known for their ability to produce detailed and contextually relevant text on a wide range of topics. These technologies find application in fields such as digital art, 3D modeling, music creation, creative writing, and product design, leveraging their unique ability to generate new creative content.
…did you say GPT?
A GPT (Generative Pre-trained Transformer) is a type of artificial intelligence model designed to generate text automatically.
GPT models are “pre-trained” on vast textual datasets (such as books, articles, and websites) to understand and generate human language. During pre-training, the model learns to predict the next word in a sentence, gaining a thorough understanding of language structure and context. This pre-training phase allows the GPT to have a broad general knowledge and the ability to generate texts that sound natural and coherent.
After pre-training, GPT models can be “refined” or further trained on specific datasets for particular tasks, such as answering questions, translating languages, or creating content in specific writing styles. This makes GPTs extremely versatile and suitable for a wide range of applications in the NLP field.
Here we are at ChatGPT
ChatGPT is an advanced language model based on the GPT architecture, developed by OpenAI. OpenAI is an artificial intelligence (AI) research organization established in December 2015 founded by a group of high-profile entrepreneurs and researchers, including Elon Musk, Sam Altman, Greg Brockman, Ilya Sutskever, Wojciech Zaremba and many others. Initially established as a nonprofit entity, the organization later evolved into a mixed “capped-profit” structure with the creation of OpenAI LP, a limited liability company controlled by OpenAI Inc. the original nonprofit entity
The “ChatGPT” artificial intelligence model was specifically designed and refined to produce textual responses in a conversational context, making it particularly suitable for applications such as chatbots or virtual assistants. Its design aims to exploit the advanced capabilities of GPT models in understanding and generating natural language, aiming to provide a conversational interaction experience that is as smooth and consistent as possible.
Strengths of ChatGPT
Natural Conversations: One of the distinguishing features of ChatGPT is its ability to understand and answer a wide range of questions (‘prompts’) in a way that feels as natural and human as possible. This aspect is crucial to creating an enjoyable user experience and interactions that seem less mechanical and closer to a real conversation with a human being.
Contextual Learning: Through the use of transformer technology, ChatGPT has the ability to take into account the context in which a conversation takes place. This means it can provide responses that are not only accurate but also relevant and consistent with the flow of the conversation, demonstrating a sophisticated understanding of the dialogue taking place.
Versatility: ChatGPT stands out for its versatility, being able to adapt to a wide range of applications. This includes customer support, where it can assist users by answering questions quickly and effectively; educational assistance, providing explanations or help in learning contexts; and entertainment, where it can participate in light conversations or help create creative content. Its flexibility makes it a valuable tool in multiple areas, from business intelligence to personal assistance.
ChatGPT finds application in numerous areas, offering automated but natural interaction with users. For example, it can be employed as a customer service tool, providing quick and informative answers to common questions, or it can be used in educational settings to help with learning and training. Its ability to generate natural language responses also makes it a useful tool for creating dialogues in scripts or for generating creative textual content.
Artificial Intelligence and Business Analysis
Now that we know what an algorithm is, what an intelligence is, and especially what generative AI is, we can analyze how these tools apply to Business Analysis. While traditional AI deals with analysis and in automation, generative AI opens up new possibilities in creative problem solving and scenario planning. Both can significantly improve a business analyst’s capabilities, making processes more efficient and decisions more informed.
Artificial Intelligence in Business Analysis
- Data Analysis and Interpretation: Traditional artificial intelligence revolutionizes data analysis. It can examine large data sets, revealing hidden patterns and trends. These findings are not always obvious to the human eye. With AI, analysis becomes deeper and more accurate. Thus, business decision-making processes improve dramatically. Companies can make data-driven decisions with greater confidence.
- Process Automation: AI also transforms business analysis through automation. It can handle repetitive tasks that would otherwise be time-consuming. These include data entry, creating standardized reports and updating systems. This automation frees up valuable time for analysts. It allows them to focus on more strategic and creative tasks.
- Predictive Analysis: AI algorithms are powerful tools for predictive analysis. They predict future trends based on historical data. This capability is critical for business planning and strategy. It helps companies anticipate market changes and manage risks. It also provides a solid basis for future decisions. With predictive analytics, companies can plan with greater confidence and accuracy.
Generative Artificial Intelligence in Business Analysis
- Scenario Generation and Simulation: Generative AI is critical in business planning. It can create realistic simulations and various scenarios. It generates data sets that mimic possible futures. This helps in planning and strategic decisions. Simulations provide a clear view of various outcomes. Thus, companies can better prepare for the future.
- Automatic Requirements Generation: In complex projects, generative AI is a valuable tool. It helps draft requirements documents. It collects and synthesizes information from a variety of sources. It ensures that requirements are complete and consistent. This reduces errors and improves project quality.
- Innovative Problem Solving: Generative AI offers unique solutions to business problems. It combines elements from different past cases and contexts. This leads to innovative and creative strategies. This approach opens up new ways to overcome business challenges. The solutions generated can be revolutionary and impactful.
Let’s engage ChatGPT itself on the topic of Artificial Intelligence and Business Analysis
As explained in another Blog article, the use of AIs is not without risk if not used consciously. On this point, I asked for an opinion — from the person directly involved.
ChatGPT, puoi scrivermi un articolo su come NON usarti per fare Business Analysis?
Sure, I will be happy to help you write an article on how NOT to use tools like ChatGPT for business analysis. The goal here is to clarify the limitations and inadequacies of these tools in this specific context so that you can better understand how and when to employ them effectively.
Title: The Limits of Artificial Intelligence and Business Analysis: A Practical Guide
Introduction
In the age of artificial intelligence and automation, tools such as ChatGPT have found applications in numerous industries. However, it is critical to understand their limitations, especially in the field of business analysis. In this article, we will explore why ChatGPT is not the ideal tool for business analysis and provide guidelines on how to employ it properly.
Lack of Company-Specific Customization
ChatGPT, being a generic artificial intelligence model, is not customized for the specific needs and context of an individual company. Business analysis requires a deep understanding of the peculiarities of the organization, which a generalist AI model cannot provide.
Absence of Access to Internal Data
Another significant limitation of ChatGPT is the inability to access internal company data. Business analysis relies on the interpretation of company-specific data, such as sales, logistics, and customer feedback. Without access to this data, ChatGPT cannot provide accurate analysis or meaningful insights.
Inability to Provide Accurate Predictive Analysis
Although ChatGPT can generate responses based on trends and historical data, it cannot provide accurate predictive analysis. Business analysis often requires the ability to predict future trends and make predictions based on complex data sets, a skill beyond the reach of ChatGPT.
Limitations in Respecting Privacy and Data Security
Data security is paramount in businesses. Using ChatGPT to discuss sensitive or private information may pose a security risk because it cannot guarantee the confidentiality or protection of corporate data.
Lack of Industry-Specific Skills
Each industry has its own specifics, regulations, and challenges. ChatGPT, not being specialized in a particular industry, may lack critical skills needed for effective business analysis in specific contexts.
Conclusion
While tools such as ChatGPT can be useful in many scenarios, it is essential to recognize their limitations in business analysis. Deep understanding of the industry, access to company-specific data, and the ability to provide accurate predictive analytics are key aspects that ChatGPT cannot meet. Therefore, it is advisable to use these tools with caution and as a complement to human expertise, rather than as stand-alone solutions.
This article aims to provide a clear and straightforward overview, avoiding metaphors, to make the concept accessible to an informed but not necessarily expert audience in Artificial Intelligence and Business Analysis.
