GenAI

Harness the full power of Generative AI | Arxus

blog

How do you integrate GenAI into your modern workplace?

Generative AI. There's a good chance you've already heard of it or worked with it. Chat GPT, the most famous example among generative AI tools, reached the milestone of 100 million users in just two months. For comparison, TikTok and Instagram took nine and 30 months respectively. “Write a comprehensive, integrating blog about generative AI with the main benefits, dangers, and how to integrate it into your workplace.” Below is the result of the prompt that our own expert, Kevin Meuldermans, tackled.

Thu, 7 December 2023

Kevin Meuldermans

Data & AI Engineer, Arxus

A great number of companies have only just thoroughly reformed their work rules and habits when we already have to learn to ride a new wave of innovation. Artificial Intelligence (AI), the technology with which we try to emulate human behavior and intelligence, is sweeping through just about all sectors at breakneck speed and is becoming indispensable in the workplace of tomorrow.

Yet Artificial Intelligence is certainly nothing new under the sun. Speech and facial recognition are examples that many of us have been confronted with for years through our smartphones. Today, the focus is more than ever on generative AI, a subdomain within AI technology with which companies like Microsoft and OpenAI want to change our way of working forever.

What is Generative AI?

Microsoft compares generative AI to a digital assistant that can solve problems quickly, efficiently, smartly, and accurately when you give it clear instructions. Contrary to what some may believe, AI is not truly alive and does not make plans to take over the world.

Through the use of Large Language Models (LLM’s), generative AI is very proficient in language and can (quickly) produce new content that closely resembles solutions made by humans. So, you write prompts or commands in 'human language' which AI tools such as ChatGPT and Bing Chat (Enterprise) take on and execute as well as possible. A digital assistant that, for example, writes documents for us, analyzes and summarizes data, or creates (unique) images, is no longer a thing of the future.

How is GenAI transforming the digital workplace?

While your generative AI assistant is writing a report on your Excel data, you can already invest your precious time in other tasks. Need inspiration for writing texts or brainstorming new ideas for a project? No problem. Moreover, ChatGPT has also proven that it can write scripts or pieces of code and thus solve problems that are outside your technical comfort zone. However, time savings, inspiration, and untapped knowledge are not in themselves the greatest strength of generative AI.

Above we briefly discussed prompts, a command in our (own) language that we can use to set a digital assistant to work. To fully harness the potential of generative AI, you may consider prompts and output as communicating vessels. The formulation, quality, and accuracy of your prompts determine how good your final output will be. “Write a blog about AI”, and, “Write an inspirational blog about AI that clearly outlines the latest developments and how they impact the workplace”, will therefore yield two completely different results.

Prompt engineering is largely what generative AI revolves around, and stands in stark contrast to the way we have been translating our search intentions to our devices for years. Thus, our search queries and commands that we assign to our smartphone, laptop, or other digital devices are today primarily based on keywords. While their use will remain relevant for search queries, it will mainly be about helping employees write efficient prompts if they want to ride the wave of generative AI.

What are the dangers and flaws of GenAI?

It's clear that generative AI offers a lot of possibilities. But what are the flaws? Let's first talk about security, the elephant in the room. Microsoft and OpenAI are busy with the development of generative AI applications that are suitable for the work context such as Microsoft 365 Copilot and Bing Chat Enterprise, but many other applications are not (yet). So be careful with what input you provide to an AI model, and always check where your input data is stored and whether it is used to train the AI model.

Additionally, generative AI is highly dependent on the input it has available, both from the data with which the model is trained and the quality of prompts. If either of these is not of high quality, a generative AI application might provide an answer that is incorrect or partially fabricated, yet seems accurate. This is also referred to as 'hallucinations'. For instance, ChatGPT is trained on data up to 2021 and therefore lacks about 2 years of information. So if you ask a current question, the model indicates it has no answer, or provides an answer that is likely to be correct or seems to be correct.

It is therefore important to analyze the results critically enough and filter out hallucinations as much as possible.

Integrate GenAI in 1-2-3

Popular applications like ChatGPT are already widely used. However, the tone changes when we want to use tools like Bing Chat Enterprise or Microsoft 365 Copilot soon with our own company data. To prepare your business for this, it is best to pay attention and devote time to the following matters:

1. Evaluate your data

We cannot emphasize this enough: the quality of your data is directly connected to the effectiveness of a generative AI model. In the context of your business, it mainly has to do with the structure and organization of your data and the content accuracy. Let's illustrate this with an example:

Your company is a producer of various basic products such as milk, water, grain, etc. As a sales manager, you want to make a proposal for a potential client and you ask your digital assistant for more information to create an appropriate proposal for your prospect. If your data is scattered across different departments that do not share their data thoughtfully with each other, you will always miss a piece of the puzzle when you are looking for information. If the 'milk' department does not share its data, you cannot find out that the customer in question may already be purchasing milk from a competitor.

The same applies to the accuracy of the data. Different datasets can cause confusion and yield incorrect results. If various departments or documents list different contacts for a client, it's difficult to determine which contact is the correct one.

2. Help discover AI

Naast de ordening van je data is het uiteraard nuttig om je werknemers aan te leren hoe ze met generative AI aan de slag moeten gaan. Hoewel velen onder ons ondertussen al geëxperimenteerd hebben, haalden we hierboven al aan dat prompts schrijven een kunst op zich is. Een workshop of cursus prompts schrijven of succesverhalen op basis van AI delen, is dus zeker geen slecht idee.

Create an environment that encourages learning and especially awareness of what AI is, how it works, and what it can do. In addition to focusing on the positive, it is also important to teach your employees to handle (generative) AI responsibly and to emphasize the flaws and dangers of the technology.

3. Consult an expert

With the right sparring partner by your side, you can truly get the most out of GenAI. An experienced expert can assist you and your company in correctly implementing this innovative technology and avoiding pitfalls. They can also discuss possibilities with you, develop use-cases, and create the most suitable AI strategy for your business.

Want to debate about Generative AI?

Get the most out of Generative AI with the right support | Arxus

Want to read more? Check these out!

blog

Microsoft Copilot is finally here | Arxus

Thu 9 Nov 23

Microsoft 365 Copilot launches a new era of personal computing

"We believe that Copilot will fundamentally change our (work) relationship with technology, and usher in a new era of personal computing," says none other than Microsoft CEO Satya Nadella. Arxus also anticipated the arrival of Microsoft 365 Copilot for some time. As part of the early access program, we have been able to test the tool intensively and in depth. Will November 1st forever be remembered as the day that Microsoft 365 Copilot first saw the light of day? Let's discuss the basics of the tool and help you prepare for the next big thing.