Generative AI in Business: Value and Approximation
If it were easy to replicate a human brain, we would be too stupid to do it.
This is how Professor Roberto Tagliaferri began his course on neural networks at the University of Salerno (or rather, the course on Cybernetics and Information Theory), encouraging us to explore the field of Artificial Intelligence with humility and determination.
It was clear to everyone that soon our lives (data + relationships) would be managed by AI algorithms, and indeed this is what slowly happened, to the advantage of a few companies capable of turning computing into revenue.
More fascinating, and perhaps less predictable, was the idea of a future where every single connected human being could exploit the potential of AI for their activities (AI Democratization).
This vision was realized in just 12 months thanks to the launch of ChatGPT (Generative AI) and for the first time a technological revolution of such magnitude was so widely accessible: 100 million users reached in just 6 months.
1. Generative AI: What’s Happening in Businesses?
The democratization of AI has introduced a decentralized transition in businesses from the bottom up. Instead of expensive technologies managed hierarchically (by assignment), businesses are now using a series of tools (by subscription) that can be used independently by each individual (BYO-AI). These tools can create value by generating advanced content, boosting efficiency and productivity by automating repetitive tasks, and generating completely new products, services, and business models.
However, the success of this transition depends on both individual responsibility and careful governance by managers. The challenge becomes how to exploit the value of generative AI while managing the risks (in particular, reputational risks).
Like traditional AI, generative AI raises concerns about privacy and ethical risks, such as hidden biases in training data. There are also new risks, such as the risk of violating copyright, trademark, and patent-protected materials using the collected data.
One final consideration is related to the “propensity to hallucinate” of generative AI, or the propensity to generate inaccurate information, expressing it in a way that appears so natural and authoritative that it makes it difficult to detect inaccuracies.
2. Systemic Value and Uncontrolled Spread of Approximations
The unanimous consensus regarding the opportunities that Generative Artificial Intelligence is unlocking in all sectors is corroborated by the generation of systemic value, produced by new ways in which people, work, and technology are merging and that is spreading in a decentralized way from various areas of the same company and between different companies but connected and interdependent.
The propagation of systemic value in a company (and between companies)
refers to the way in which the value generated by an action or
initiative spreads and contributes positively to different aspects
of the organization, transcending the boundaries of the single unit
and propagating to all the nodes (companies) of a network.The uncontrolled spread of approximations in the use of generative AI can compromise the overall health of the company (approximations can accumulate over time, affecting the overall quality of business decisions), undermining its reputation, reducing the value generated, and negatively impacting its operational performance.
3. Addressing it the right way
Given the speed of innovation in AI, it is difficult to identify a single framework that can be used in every company and for every scenario. In its paper “Building a Foundation for AI Success — A Leader’s Guide,” Microsoft AI identifies five pillars for a successful AI project in the business context. The first critical factor for success is not technical in nature, but is related to the organization and corporate culture, of which training is an integral part.
The decentralized nature with which generative AI is spreading in companies makes it difficult to govern its use from scratch. Therefore, the starting point is necessarily adequate and necessary training at all levels, essential to ensure that company personnel have the skills necessary to generate and correctly interpret the results from generative AI and understand its potential and limitations.
Of course, strategic objectives drive adoption. “Rather than starting by asking ourselves what AI can do, we need to turn the telescope around and ask ourselves: ‘What are you trying to do in your company and how can AI help?’” Jason Price, Director of Specialist Management at Microsoft.
Finally, managers in particular have the task of developing a solid governance to ensure responsible use of generative AI, introducing processes of monitoring and continuous evaluation, transparency in decision-making processes, and defining and disseminating clear ethical guidelines and accountability.
4. Conclusion
Starting from Professor Tagliaferri’s statement (although the quote is actually from Marvin Minsky), it is evident that we have not yet managed to replicate a human brain. It is important not to fall into the trap of deluding ourselves that we have succeeded and, therefore, to take the hallucinations of AI as true. Nevertheless, it is worth reformulating:
If it were easy to replace tasks in our work with artificial intelligence, we would be too foolish not to do it.
…someone else will do it in our place.
(V. Riccardi)
References
- A) Four essential questions for boards to ask about generative AI — QuantumBlack, AI by McKinsey — 2023 (Link)
- B) The Generative AI Dossier — A selection of high-impact use cases across six major industries Deloitte Consulting (Deloitte AI Institute) — 2023
- C) Now decides next: Insights from the leading edge of generative AI adoption Deloitte Consulting — Jan-2024
- D) Building a Foundation for AI Success: A Leader’s Guide — Sep-2023 Microsoft AI
- E) The Emotion Machine: Commonsense Thinking, Artificial Intelligence, and the Future of the Human Mind”, Marvin Minsky — Simon & Schuster, 1986, p. 239
