What impact will ChatGPT have on typical business intelligence processes?

About every ten years, there is a profound technological change that is able to steer central business processes in completely new directions. In the last few decades, these have undoubtedly included the development and strong advancement of Google in the late 1990s and 2000s and the enormous growth of mobile technologies in the 2010s. Currently, another game changer is emerging in the form of tools with generative AI, first and foremost ChatGPT. Generative AI is likely to have a strong impact on business intelligence (BI) practices.

Generative AI offers new opportunities to increase efficiency

Artificial intelligence (AI) has long been an important factor in increasing efficiency in many business processes. However, generative AI takes AI possibilities to a whole new level with its less analytical and more manufacturing approach.

The most obvious business uses for ChatGPT (or similar tools) are in supporting (potential) customers through live chat integration and in writing copy for websites, emails, flyers or similar advertising or marketing media. However, such systems can also contribute to business success in a much more differentiated way.

ChatGPT is able to analyse large amounts of data and even process it independently. This includes customer information, financial figures, competitive data and other values that enable those responsible to better understand the behaviour of their buyers, market trends or competitors and to align their own companies accordingly. This makes the technology extremely interesting for typical business intelligence (BI) processes.

What does generative artificial intelligence mean in the context of BI?

The term "generative AI" covers all types of automated or algorithm-driven processes whose goal is to generate data autonomously.

Machine learning approaches form the basis of all this. With the help of corresponding technologies, huge amounts of information, which often come largely from the web, are used as the basis for generated answers or data products. This is done with the help of algorithms that have been developed beforehand precisely so that they can distinguish, analyse and process certain things.

The decisive difference in this form of AI is the generative properties that give it its name. This means that such systems can largely create something new on their own. In contrast to this is discriminative AI, which has so far been commonly used in systems to increase efficiency for business processes. It pursues a more analytical basic idea.

Such procedures are also used in business intelligence. Traditional BI describes how a company has performed today and/or in the past. The focus is usually on answering the questions "what happened?" and "what should change?". However, why something happened and what the next steps should be are not addressed.

This can change fundamentally through the use of generative AI. In general, highly agile processes can be achieved through appropriate technologies. In the future, AI-generated ideas for the data-based optimisation of business processes may only need to be approved or thought through further.

ChatGPT can revolutionize data analysis and decision-making in BI

Due to the generative nature of ChatGPT technology, companies can not only analyse data very effectively, but also have questions answered around the analyses and related potentially important business decisions. Corresponding business intelligence processes are thus no longer purely descriptive analyses. They are evolving into agile operations based on natural language, in which generative AI can exert enormous influence in a variety of ways. Ideally, it is integrated into a proprietary BI tool. This ultimately enables companies to obtain more flexible and deeper insights with less effort within a centralised system.

ChatGPT can interpret questions in natural language and then use machine learning to access, capture and analyse the required data and ultimately return an accurate answer in natural language. Manual data extraction and analysis are no longer an issue. Companies get faster, more accurate and more understandable insights for all stakeholders.

Generative AI can be particularly helpful to companies in the following BI areas.

• Data capture: ChatGPT BI integrations allow managers to capture data from multiple sources in an instant. These include databases, APIs or spreadsheets. This makes it much easier for companies to centralise, organise and process information.
• Data cleansing and preparation: ChatGPT AI can provide significant support for data cleansing and preparation. The technology can be used to remove duplicates, inconsistencies or errors that can severely affect the accuracy of data analysis in real time and in a fully automated way.
• Data visualisation: Using the generative ChatGPT AI, those responsible can have individual and interactive visualisations created without any intervention. Such visualisations enable a simpler understanding of the relevant data and a more targeted interpretation and final evaluation.
• Predictive analyses: ChatGPT-BI can be used to carry out predictive analyses and provide an outlook on future trends. The ability to ask the system questions in natural language and receive specific answers simplifies processes immensely. Making data-driven decisions thus becomes more efficient and potentially more effective overall.
• Reporting: ChatGPT is able to generate specific reports in various formats. The system can output PDFs, tables or presentations. This makes it much easier for companies to share and collaborate on data insights with all stakeholders.

Why is it important to use ValueWorks or other BI software in this context

A lot of these features sound very promising, but as data specialist we also know that these things are nice to read on paper, but will hardly be implementable in reality. Reason for that are three things.

1. Semantic meaning of data: Given the situation that you would dump data into ChatGPT, there is a high probability that ChatGPT could do statistical analyses on the data set itself, but be unable to uncover the semantic meaning of the KPI. That said the lack of context would not let it be interpreted in a larger context like relations to other, external KPIs.
2. Data preparation: While ChatGPT will be able to support things like data cleansing, what it will not do is help with basic enrichment of data to be able to automatically calculate KPIs. For instance a deeper knowledge of systems like Hubspot is necessary, especially how it needs to be set up.
3. Data privacy: You don't want to load detailed company and personnel data in ChatGPT. You just don't want to.

Therefore, you will still need to work with a BI tool upfront for data preparation and to ensure that data privacy is adhered to. For the first part, ValueWorks automatically creates semantic classification and helps with data preparation. In addition, ValueWorks adheres to the highest professional standards and it has guardrails in place so you don't get into GDPR-related trouble.

Final note: caution is advised

The results of generative AI applications are often impressive. This gives the impression to many that it is a mature technology that can be used immediately and without further ado in business intelligence. Unfortunately, this is not the case.

It is true that generative AI can already achieve quite a lot. However, the still young technology demands great caution from those responsible for BI in companies. Many fine details still need to be worked on in order to finally obtain a system that works independently and with a high degree of security.

At present, the BI potentials are mainly diminished by the following factors.

• Generative AI can error even intentionally produce incorrect information: It is relatively often observed that ChatGPT completely "self-consciously" generates inaccurate information as answers to user questions. Thus, incorrect data may be arbitrarily included in texts or the system may try to cover up its ignorance (yes, ChatGPT intelligence is not omniscient) with inaccurate statements. Up to now, there is no built-in mechanism that completely excludes such processes or informs users about careless interpretations or generally questions the results. In the context of business intelligence and resulting far-reaching decisions, such behaviour can of course not only be annoying, but even highly dangerous. The ChatGPT AI must not be trusted blindly.
• Insufficient filters to detect inappropriate outcomes: Generative AI can make a great many appropriate decisions on its own. However, there are relatively frequent cases where inputs are not channelled correctly. In the context of business intelligence, this can mean that information is basically correct but interpreted in an undesirable direction.
• Lack of up-to-dateness: The ChatGPT AI bases its processes largely on data from the past. This is not up-to-date on a daily basis. This means that even if the system is fed with currently relevant business information, it is possible that the artificial intelligence interprets this on the basis of outdated data and thus only produces insights that are of limited use. Such systemic biases are currently being addressed. However, it will probably take some time before the AI reliably circumvents them.
• Individual requirements cannot be automatically recognised and accepted: Managers cannot rely on ChatGPT AI to interpret their BI requests correctly on its own. Companies must adapt or feed the technology so that it correctly implements their intentions. This point is also de facto relevant to the potential problems mentioned earlier. The more precisely you tell generative AI what to do, the more useful the results tend to be. The creation of appropriate conditions requires expertise and other specific resources. Both are only available in very few companies, which is why it usually makes sense to bring an experienced partner on board.

Sources:

https://www.unite.ai/generative-vs-discriminative-machine-learning-models/

https://www.sap.com/germany/insights/what-is-business-intelligence-bi.html

https://www.computerwoche.de/a/was-ist-generative-ai,3614061

https://www.quinnox.com/chatgpt-the-new-business-intelligence-tool-to-make-informed-decisions/

https://aineox.com/en/unlocking-business-intelligence-and-analyzing-data-with-chatgpt/

https://ts2.space/en/chatgpt-and-its-potential-in-enhancing-business-intelligence-and-data-analysis/

https://puiij.com/index.php/research/article/view/11

#AI #BusinessIntelligence #ChatGPT #DataScience #MachineLearning #NLP #NaturalLanguageProcessing #BigData #Automation #DigitalTransformation #Innovation #Technology #ArtificialIntelligence