AI in business: Beyond the chit-chat
With the November 2022 introduction of ChatGPT, artificial intelligence (AI) chatbots are now part of our everyday work and personal lives. But are chatbots really the best use case for implementing AI in business?
AI in business – short & sweet
- Chatbots using Large Language Models (LLM) can be very useful but still have significant limits
- Business use of chatbots comes with the risk of disclosing sensitive company information
- A more promising business use is in endpoint protection, e.g., in the form of anomaly detection
- The focus should be placed more on the practical benefits of AI in business – beyond “chatty AI”
Using AI in business correctly: Opportunities & risks
Recently, I have been reviewing discussions about the benefits of AI for our private and professional lives. Since its introduction in November 2022, countless articles have focused on
Large Language Model (LLM)-based cloud service ChatGPT from OpenAI. Fortune magazine even called its introduction the "iPhone moment" of AI. Since then, other LLM-based
chatbots such as Google’s Gemini, Anthropic’s Claude and Microsoft Copilot also have been widely adopted for personal and business purposes.
Discussions about the use of LLMs and chatbots often are divided into two camps:
- Some praise the time savings and ease of use and cite many examples.
- Others focus on the shortcomings of the technology such as “hallucinations” or its tendency to invent “facts”. Critics also point to multiple examples of blatant errors in LLM model reasoning and calculations.
How is it possible that an AI can produce well-written essays on complex topics but fail at simple math problems? The reason can be found in the term "chatbot" itself; ChatGPT is neither a search engine nor an encyclopedia, neither a history book nor a calculator. It is, as its name implies, a "chatting robot".
AI & IT security: Why it takes more than just ChatGPT
ChatGPT is trained to make conversation in response to questions or prompts. In fact, the chatbot performs this task surprisingly well. Its responses seem like they come
from a friendly, very talkative and highly educated person, as if it were someone sitting next to you on the train with whom you spontaneously struck up a conversation.
Chatbots share another trait with that fellow passenger – nobody likes to admit they don't know something. The result is that while chatbot responses may contain a lot
of truth, they also may include some half-truths and occasionally false or inaccurate information along the lines of, "I heard somewhere that…" But when the passengers
arrive at their destination they've at least had a nice chat.
Protect Endpoints – Smart and Secure
Effective endpoint management is essential for strengthening your organization's cybersecurity. By strategically deploying the right security solutions,
you can reliably protect devices from threats and detect vulnerabilities early. Our free white paper on vulnerability management reveals how intelligent strategies and
technologies can significantly enhance your IT security.
Download now for free!
Using ChatGPT in business – with caution
At work, it’s possible to apply a chatbot’s eagerness to communicate and tame its tendency to play a little fast and loose with the truth. A simple example is having employees use a chatbot
service to draft individual responses to customer inquiries. That way, users only have to check and modify the chatbot’s output instead of taking time to create each response from
scratch.
But be careful: chatbots are a cloud services, so users should take care not to include
sensitive internal company information in queries, prompts and replies!
A more complex option is for companies to train their own LLM. Alternatively, they could restrict a basic model to source information only from trusted predefined data sources using a
process called RAG (Retrieval Augmented Generation). Both methods help limit the amount of false information generated by the chatbot
More useful approaches for business AI
Nevertheless, it is a little surprising that ChatGPT and other chatbots grab so much attention when discussing the benefits of AI for business purposes. After all, there are other
types of AI that can be just as useful. Some of them, such as common speech recognition tools, have been in use for a long time without generating the hullabaloo
that ChatGPT did 2+ years ago.
AI has also long been used "under the hood" in endpoint management. Endpoint security solutions such as
Microsoft Defender Antivirus, for example, no longer just use comparisons with known virus signatures to protect end devices. Instead, they use machine learning (ML) to detect anomalies.
This is because ML is very good at using statistical analysis to learn the "normal behavior" of network traffic in huge volumes of data. This enables endpoint security software to
detect outliers and statistically unlikely clusters, e.g., an endpoint device whose behavior suddenly departs from its usual pattern in a suspicious way.
AI-based security solutions in practice
In such cases, the AI- or ML-supported security solution sounds the alarm. The company's IT team can then manage the end device in question using the baramundi Management Suite or other suitable solution. In addition, baramundi is a comprehensive UEM solution that offers extensive software update management, data encryption, interface control and vulnerability scanning. All in all, the combination of UEM and a solid AI-supported endpoint security approach comes with an added benefit: it won’t start spreading bogus information in chats with users.
Conclusion: AI in business – focus on benefits instead of hype
While tools like ChatGPT can add value in the workplace, companies should take a strategic approach to where AI truly delivers impact —such as in cybersecurity or process automation.
Not every chatbot is a smart solution—but when AI is thoughtfully integrated, it can reduce risk and boost efficiency.
