Generative Artificial Intelligence: A New Form of Alchemy?

Since the launch of Midjourney, followed by that of ChatGPT, I’ve spent numerous hours using, exploring, and understanding generative Artificial Intelligence, while piloting several projects based on this technology. My interactions with clients, colleagues, and friends have been punctuated by incessant questions: “What exactly is this generative AI?”, “What’s its real purpose?”, “Is it just a passing trend?”, and a major concern: “Will we be replaced by machines?”. In this article, I strive to offer some insights — while avoiding technical details — demonstrating how generative AI acts as a transformative force marking the beginning of a new era.

Please note: the opinions expressed in this post are strictly my own and do not necessarily reflect those of EY.

The recent decades have witnessed a metamorphosis in our access to and relationship with knowledge and culture. The advent of the computer marked a turning point in our ability to process and analyze vast quantities of data, paving the way for the information age. With the emergence of the Internet, the reach of this data expanded, allowing everyone, no matter where they are, access to a wealth of knowledge once reserved for the elite. Social media further accelerated this democratization, giving us a platform to share, debate, and exchange ideas on a global scale. Lastly, the smartphone made this technological revolution omnipresent, putting the world of knowledge at our fingertips, accessible anytime and anywhere. In the wake of this technological evolution, generative Artificial Intelligence represents an additional step in our relentless quest for understanding and enhancing our cognitive abilities. It shines a light on our limitations as much as our triumphs, our past as much as our future.

Rooted in our heritage, generative AI mainly relies on human data. This data, be it textual, visual, or otherwise, is the result of years of knowledge, experience, and culture. It directly reflects our history, emotions, successes, and failures. By using this knowledge, generative AI isn’t just a technological tool but also a living archive. However, it’s crucial to recognize that this archive isn’t without flaws. Data, often bearing the biases and prejudices of their time, can influence how generative AI perceives and interprets the world. These biases might be unintentional, reflecting past or present social norms, but they still raise serious ethical questions. How can we make certain that generative AI is fair, just, and doesn’t perpetuate injustices? How do we ensure its interpretation of the past doesn’t shape a biased future? Furthermore, legitimate questions arise concerning data sources and copyrights. Who really owns what AI generates? How do we distinguish between inspiration and copying, especially when AI can replicate styles and ideas with such fidelity? These concerns, though important, don’t limit the impressive potential of this technology. Indeed, beyond the rich legacy it harnesses, generative AI pushes us to the brink of a new era. An era where its ability to learn, interpret, and innovate goes beyond even the vast wealth of human knowledge it’s absorbed.

Generative AI relies on complex deep learning structures, encompassing billions of parameters, to produce a myriad of results on demand. Far from just replicating what it’s “seen”, it creates original content with such credibility that it could be mistaken for human production. Take ChatGPT as the primary example of this prowess, as this service introduced the general public and media to generative AI in late 2022. This chatbot can capture the essence of a 19th-century poet, a contemporary physicist, or a film critic to provide detailed analyses, write inspired texts, answer complex questions, and engage in enriching discussions with users worldwide. But generative AI doesn’t excel only in textual production. With tools such as Midjourney, Dall-E or Gen-2, the world of imagery and video is also shaken up. Midjourney, for instance, can simulate art styles from the Renaissance to Contemporary Art, producing images and designs with a level of detail and finesse rivaling human professionals. Its potential isn’t limited to replication and extends to innovation, producing visuals that sometimes offer disruptive aesthetics. Generative AI combines knowledge, techniques, and skills at an unmatched speed and scale. Where a group of human experts might take days to conceptualize, discuss, and finalize a project, these systems do it in a fraction of that time, always iterating and optimizing.

Generative AI isn’t just a powerful tool; it opens a new chapter in our relationship with technology. It goes beyond simple interaction to establish a seamless dialogue between humans and machines, thanks to its unique ability to comprehend natural language. It’s not just a query being made, but a real modern incantation — a “prompt” — that guides the AI towards a creation or solution. In this partnership, AI doesn’t just mechanically respond; it assimilates feedback, adapts, and enriches the creative process. Take the example of Revolve, a high-end fashion e-commerce platform. In an innovative move, the brand’s designers collaborated closely with AI technology for design creation. After providing initial sketches and instructions to the AI, it returned several proposals. Guided by the designers’ feedback, the AI refined its ideas, ending in an avant-garde collection. The pinnacle of this synergy was the creation of a leather jacket that quickly caught customers’ attention. Although the initial inspiration came from the AI, each jacket was meticulously hand-painted, symbolizing the harmonious fusion of technological innovation and traditional craftsmanship. Nonetheless, despite its impressive advancements, generative AI has its limits. It’s still dependent on the data it relies on and can lack human nuance, creativity, common sense, and real-world knowledge. For instance, in architecture, a generative AI might design an aesthetically impressive building, but one that’s completely impractical. It might suggest a design with windows placed at unreachable heights or rooms without access, simply because it’s optimized for maximizing space or light without considering functionality or human experience. Such shortcomings clearly show that AI, while incredibly powerful, still needs human oversight and direction to ensure practical and relevant results. The fear of humans being replaced by machines is thus quite exaggerated. Instead of a threat, generative AI should be seen as an ally in our quest for innovation.

Through this new form of alchemy that is generative AI, we’re not turning lead into gold, but raw data and accumulated knowledge into fresh ideas and perspectives. Every individual, equipped with this technology, can go beyond their limits and reach higher levels of creativity and ingenuity. This is further confirmed by a study published in early August 2023, conducted by Anil R Doshi of University College London and Oliver Hauser of the University of Exeter. The study showed that stories written with the assistance of generative AI were seen as more creative by a panel of professionals than those written unassisted. Additionally, these stories were better written and thus more enjoyable to read. This suggests that generative AI isn’t just a machine creating for us but a catalyst, amplifying and enriching our potential. Thanks to it, we’re transforming not just our environment and work methods, but also how we see our own capacity to innovate and create.