With AI, everything changes

At 7N, we had the pleasure of hosting Jan Damsgaard, AI expert and Professor at Copenhagen Business School, and Murugasan Nielsen, prompting specialist and Managing Partner at Ustrat, for an exciting evening on the topic of AI and its transformative impact on

the world. Read along here for some of the captivating insights from the two experts.

An unavoidable part of the future

AI may seem like a black box, stirring uncertainty, risk, and the threat of making us obsolete. Yet, to maintain our competitiveness – whether as individuals, businesses, countries, or entire continents - it is crucial that we overcome these apprehensions head-on and actively embrace AI.

Despite Denmark's status as one of the most digitally advanced nations, there is a discernible reluctance towards AI adoption, Jan explains. This hesitancy is mirrored across the EU and poses the risk of falling behind as AI development progresses. He underlined that rather than viewing AI solely through a regulatory lens, European countries must embrace the potential of these technologies to remain competitive. After all, technological progress will continue its rapid development elsewhere, irrespective of the level of involvement in Europe.

Jan further argues that while national and international regulation is important, its effectiveness in controlling AI tools is limited. The real governance needs to occur inside the firewall, ensuring responsible and ethical AI practices within organizations.

Will we be replaced?

If you ask Jan, AI is indeed a game changer — a transformative breakthrough surpassing the impact of the smartphone. This sentiment necessitates us to embrace it and develop the necessary competencies to utilize its potential. But can AI completely replace us? Far from it. Using the analogy of a parrot trained for specific responses, he illustrates that while AI can process vast amounts of data, it operates on probabilities and lacks true comprehension.

Rather than replacing us, AI enhances our capabilities by leveraging its access to extensive data. This enables us to concentrate on more strategic tasks while it handles administrative duties efficiently. As Jan puts it:

“You are still responsible — you are not replaced; you are enhanced.”

AI will also become a natural part of how the next generations learn and work. From elementary school to university, students are already adopting AI at such a high level that it necessitates rethinking of examinations, teaching methods, and school subjects. While one may think this will make students less knowledgeable or skilled, Jan noted that early evidence is, in fact, pointing towards a greater quality of learning because AI elevates the focus to more creative and challenging aspects of their subjects. Additionally, it empowers individuals challenged by language or learning barriers to overcome these obstacles, allowing them to excel in areas that may have otherwise been too challenging to undertake.

The shadow side

"Technology comes with a dark side - it's unavoidable - but that doesn't mean we shouldn't implement it," says Jan Damsgaard, highlighting the inevitability of challenges alongside technological innovation.

To effectively address these challenges, you have to exercise caution when using sensitive data in open-source models and remain vigilant against various types of bias. Murugasan underscores the importance of being aware of confirmation bias, highlighting how easily data can be manipulated to fit preconceived notions. It is also no secret that algorithms algorithms, such as ChatGPT, exhibit biases reflected in human-generated data. The difference, Jan points out, is that in humans, it is subconscious, but with AI tools, these biases can be consciously addressed.

Train your AI

In many ways, internal AI tools resemble office trainees – they require quality input and training to become truly valuable and unbiased. While these tools possess access to extensive data and exceptional learning capabilities, they lack contextual understanding without proper guidance and instruction. To effectively train your AI tool, you must begin with a cleansed and locked dataset before allowing access to broader data. Jan emphasizes the importance of quality over quantity in this process to avoid bias and optimize its final performance. Training involves both supervised and unsupervised methods, along with providing the AI tool with feedback on its responses. 

When it comes to open-source models, Murugasan emphasizes the importance of creativity and critical thinking for their effective responses. He notes that asking more general questions also yields more general answers. Murugasan advises users to provide context and then prompt the AI tool to elaborate on its responses to guarantee higher-quality outputs.

Find more AI articles here: https://www.7n.com/insights-overview/?tags=AI