Artificial intelligences such as GPT-4 are a cultural revolution. Since the latest iteration of OpenAI was unveiled, the examples showing it are even more impressive. GPT-4 is indeed multimodal, in other words, it is able to understand images or sounds, unlike ChatGPT. His comprehension and reasoning skills are better.
On the technological level, even if this family of AI called LLM (large language model) is currently on the rise, no one knows if this architecture will still be there in a few years. Prominent researchers like Yann Le Cun, Head of AI at Meta, believe that future AI systems will need to be fact-based, without content generation, and as such will have a very different architecture from the current generation of LLMs.
Besides, OpenAI is just one company among others alongside Cohere, Github, Hugging Face, DeepMind, not to mention Google, Baidu, Nvidia, Microsoft, Meta, Tencent or IBM. The amount of private investment in artificial intelligence has grown from $6 billion in 2013 to $176 billion in 2021, according to the Stanford AI Lab. If the moon landing is yet to come, the launch of ChatGPT was the sputnik moment of the AI race. A kind of culture shock. Everyone has heard of it. According to an Odoxa survey, half of French people have already heard ChatGPT being mentioned. A score that climbs to 86% in the sphere of technology, digital and innovation professionals.
With AI, an explosion of content in sight?
The most interesting question is that of the economic consequences of these tools. The history of innovations is intimately linked to the shift in value they trigger between different population categories and geographies. Their social acceptability has often depended on the slowness of these changes. Value displacements related to tools can be of two kinds. In some cases, they lower the barrier to entry and expand the number of people able to produce goods and services. In others, they make individuals who are already part of production circuits more productive. This is the most exciting debate about this technology. The strongest opinions abound, between those who consider that learning to code is equivalent to learning to develop film photos before the invention of the digital camera and those who believe that LLMs are formidable in the hands of coders or talented authors who know how to chisel out the basic instructions that must be given to them. In short, will Marc Levy be able to write six books a year and therefore significantly increase his income, or will thousands of people be able to access literary production, and therefore explode the depth of the shelves or lower the price ?
I lean towards the first hypothesis. Because if creation becomes a commodity, whether in literature, computer programming or consulting, there are already multiple bottlenecks in the distribution of this content. We have already experienced a content explosion with the increasing access to higher education and therefore the increase in the number of people capable of producing content or intellectual services. However, in this era of abundance, the selection was even tougher, and the concentration took place on figures identified with assured success. In this world of abundance, distributors have tended to become more conservative.
Even though communities have replaced centralized distributors, the phenomenon remains the same. A minority gets the majority of the attention and the fans. The Internet has made it possible to bring out superstars capable of reaching much larger communities or start-ups capable of addressing global markets. AI has the potential to explode these stars by empowering them to produce more. To sum up, will AI be communist, redistributing opportunities between individuals, or on the contrary capitalist, participating in the concentration of wealth in the hands of a few? If, as I fear, the second aspect prevails, expect to see the Luddites reappear soon.