Generative artificial intelligence as Archimedes’ lever

It seems that the hype surrounding generative artificial intelligence is finally dying down, and with it, the number of alarmist articles and misinformation about this new technological frontier. For instance, in recent months, several articles have been published claiming that OpenAI, the company behind ChatGPT, was headed for bankruptcy due to the daily maintenance costs of its AI models, all without citing any sources. Conversely, a recent study from The Information website estimated that OpenAI, which we should note is a partner of Microsoft, will generate over a billion dollars in the next 12 months.

In addition to OpenAI, there’s Nvidia, a leading company in the production of essential hardware for developing modern generative artificial intelligence models, setting record after record in the market. Even initially skeptical or critical entities like Apple and the controversial Elon Musk have entered this tight competition, albeit significantly behind. Alphabet (Google), on the other hand, has caught up by integrating AI into many of its services, similar to what Microsoft has already done. The Mountain View company, whose historic search engine remained unaffected despite Microsoft’s attempt to integrate GPT-3.5 into Bing, aims to develop new AI models superior to the competition.

In this scenario where AI is proving to be anything but a bubble, the collective fear that we might call the “Terminator effect” seems to have vanished. No longer are there people spending their time imagining scenarios where AIs exterminate humanity. Public opinion now seems to fall into two camps: on one side, there are the detractors and those who have tried out AIs for a short period, quickly losing interest (at least until the next big thing), while on the other side, we have those who have already integrated AI into their workflow, opting for various premium subscriptions.

Alongside the decline of catastrophic visions, there has been an increase in privacy oversight and control tools. For example, it is now possible to decide whether or not to allow OpenAI and other companies to use our material, published on social media or blogs/websites, to train their artificial intelligence models. There are also several ongoing legal cases, especially in the USA, that could clarify the main concerns regarding copyright.

Regarding more scientific or philosophical questions, such as the possibility that these GenAI (Generative Artificial Intelligence) might evolve into AGI (Artificial General Intelligence) with capabilities comparable to, or even surpassing humans (ASI – Artificial Superintelligence), potentially becoming sentient entities, there are already studies based on preliminary investigative methodologies.

Setting aside these debates, which are often misleading and based on now outdated perspectives, we need to reflect on what generative artificial intelligence truly represents today and its real-world applications. The area that most fascinates and interests me is the editorial realm, encompassing both literary and graphic works. Before delving deeper, there’s an essential clarification I wish to make, even if it might seem evident. As of now, no artificial intelligence systems can produce texts, images, or audiovisual works entirely on their own. They always require a human input or a prompt, often referred to as a ‘prompt’ (there are also systems that can dissect a broad request into sub-tasks to be executed recursively). While AIs are indeed a novel and transformative creative tool, they fundamentally remain software tools in the hands of humans. Currently, there are no instances of AIs creating texts or images autonomously or acting on their own ‘volition’.

With the assistance of artificial intelligence, one can craft social posts or articles featuring high-quality images and text in mere minutes. If the aim is to draft a short story that’s more intricate than what a language model like ChatGPT might produce from a single prompt, it could take perhaps an hour or slightly more. And for an even more in-depth and intricate composition, a few days might be necessary.

Within this progressive scale of engagement with such tools, the human influence on the final work becomes increasingly pronounced. Ultimately, an author cannot, in good conscience, claim that the entire work is solely the product of their genius. Yet, it’s natural for them to believe they can assert copyright, even though artificial intelligences have not only assisted but also amplified their capabilities, leading to significant time savings: hours of work for tasks that previously took days, and days for those that once required weeks.

Generative artificial intelligences are frequently described as the ‘new industrial revolution’ or the ‘new electricity’. Both terms aptly convey the revolutionary impact of this phenomenon. However, I’d like to introduce an even more fundamental analogy: considering these tools as the creative equivalent of Archimedes’ lever. This comparison isn’t about physical or mechanical leverage but pertains to creative fields such as writing and graphic production, which are foundational to the publishing industry.

Such advancements will certainly impact the job market. As AI enhances company productivity, we might see a reduction in workforce numbers, trending towards an optimal employee count that will, inevitably, be less than what it is today. The situation with Bild, the German tabloid that has already laid off hundreds, stands as a testament to this. On the brighter side, numerous small businesses could harness this increased efficiency, becoming more competitive and, as a result, creating new job opportunities.

Today, it’s essential to view generative artificial intelligences not as replacements for humans, but rather as tools that amplify our creative and intellectual abilities. We must move beyond the often obsessively debated ethical issues, understanding that these tools, regardless of how distinct they may seem, respond to user commands just like any other. Likewise, instead of becoming overly fixated on copyright dilemmas, we should harness these technologies to our best advantage and await definitive clarity from legislators.