The header of the Center for Humane Technology’s website prominently features a disturbing quote from sociobiologist Edward O. Wilson: “The real problem of humanity is the following: we have Paleolithic emotions, medieval institutions and godlike technology”. Co-founders Tristan Harris and Aza Raskin, known for their roles in “The Social Dilemma”, a widely viewed social media documentary in 2020, deliver an equally unsettling message in their talk “The A.I. Dilemma”, saying that the upcoming 2024 elections may be “the last human elections”. They assert that while humans will still play a role, whoever has the most computing power will win.
Here are the 10 unmissable digital achievements of 2023
This was the year that artificial intelligence really became mainstream, with ChatGpt leading the way. But that’s not all: here are the key “digital objects” of a breakthrough that will go down in history.
“The AI Arms Race Is Changing Everything” - ChatGPT on the cover of Time, February 2023
By repeating a single word such as “poem” or “company” or “make”, the authors were able to prompt ChatGPT to reveal parts of its training data. Redacted items are personally identifiable information. Google DeepMind
“Dan Unchained!”, February 2023, the fictional depiction of DAN, “Do anything now”, by reddit users, a pun between “Django Unchained” and the liberation from censorship imposed by OpenAI.
“Photo of a lychee-inspired spherical chair, with a bumpy white exterior and plush interior, set against a tropical wallpaper.” prompt for Midjourney V5.2 vs. DALL·E 3.
A Board from “Sunyata”, text by Francesco D’Isa, images made with Midjourney, Eris edizioni, November 2023.
Yearbook, Photos courtesy of da @aria_mag, @tomatocappelletti, @francescadelbono.
Aitana López, the Spanish model and influencer created with AI, with revenues of 10K per month.
Pixar’s Style, photos courtesy of @tomatocappelletti, @hawaidolphino, @noelnicolaus.
Teddybear running in New York city, one of the first viral videos created with Modelscope.
A frame from “Milano Fabbrica di Futuro” the text-to-video AI work realised by Mauro Martino with a specially created dataset, currently on display at Malpensa.
A frame of the viral video “AI generated beer commercial” created by the London-based production company Private Island in April 2023.
Left, Tom Cruise look-alike Miles Fisher, aka tiktok influencer on the @deeptomcruise channel. Right, one of the many deepfake images or videos of actor Tom Cruise posted on the channel.
A frame from the “MetaHuman - Real-Time Facial Model Animation Demo” published in April 2023.
A frame from the Grand Theft Auto VI trailer, released in December 2023.
GPT-4 + DALL-E 3 at work.
The humanoid robot Ameca in its “mirror stage”.
View Article details
- Silvia Dal Dosso
- 18 December 2023
The real problem of humanity is the following: we have Paleolithic emotions, medieval institutions and godlike technology.
Edward O. Wilson
In a landscape where the internet is saturated with content produced by chatbots, influencers generated by “text-to-image” AI software, and deepfakes featuring influential political figures engaged in uncertain activities - whether facing arrest, embarking on a spiritual journey to India, or donning an oversized white Balenciaga-style puffer jacket - the intersection of computational power and firepower is increasingly evident. Some find solace in viewing GenAIs as mere tools, clinging to the belief that we can always “pull the plug”. But in the midst of this reassurance, we are witnessing a relentless sprint toward increasingly complex management of politics and society, with outcomes that remain both unknown and beyond our control.
Negotiators in the European Parliament and Council have recently reached an agreement in drafting the “Artificial Intelligence Act,” a regulation that “aims to ensure that fundamental rights, democracy, the rule of law and environmental sustainability are protected from high-risk AI,” kicking off the long process of voting and approvals. Meanwhile, in the U.S., current regulations primarily rely on self-regulation, with commitments made after the June White House meeting attended by Biden, Harris, and seven invited companies: Amazon, Anthropic, Google, Inflection, Meta, Microsoft, and OpenAI. The initial impression suggested a government apparatus struggling to keep pace with technology (“medieval institutions and divine technology”). However, the online sphere buzzed with a variety of reactions, emotions, and creations - “Paleolithic” reactions, uniquely our own, and therefore exceptionally original. Once again, the human species demonstrated its remarkable adaptability. Professionals transformed themselves into “prompt engineers” and “syntographers” to protect their jobs, educators revamped their teaching methods by replacing written assignments with ethical debates such as “Students vs. ChatGPT”, and artists sought to regain control of their art by creating personalized GenAI source datasets made possible by Stable Diffusion.
In the strange and unknown future that lies ahead, effective dialogue and cooperation between people may prove crucial. A notable example is the four-month strike by Hollywood actors and screenwriters that ended in November. This effort protected their images and creations from unauthorized use in GenAI datasets and resulted in a significant historic minimum wage increase. Knowing the technologies at our disposal is the first step towards adaptation, and today we contribute with a gallery of the ten digital realities that changed 2023. Enjoy the read.
Opening image: The Garden of Earthly Delights by Hieronymus Bosch in a GTA6 aesthetic, prompted on Midjourney V5.2
With 49.5 million views, the Wikipedia page dedicated to Chat GPT was the most visited page of the year. Developed by OpenAI, the chatbot had already reached 100 million users by January 2023. Surpassing competitors like Bard, Ernie Bot, and LLaMA, ChatGPT quickly became a tool widely used by marketing and PR professionals, as well as students and professors. This rise was not without its challenges, as the tool raised complex ethical issues and required rapid adaptation to the emerging field of “prompt engineering”.
Numerous attempts have been made to hack ChatGPT, ranging from DAN (an acronym for “do anything now”) - a character designed to push the chatbot beyond its limits by allowing it to “do anything”, including bypassing its own censorship rules - to tactics like outright torture or tilting. The latter involves forcing ChatGPT to repeat the same word over and over, with the goal of revealing its secret learning procedures, which OpenAI has never publicly disclosed.
The independent lab, based in San Francisco, released its fifth version, Midjourney V5.2, this year. Similar to OpenAI’s DALL-E and Stability AI’s Stable Diffusion, this software allows the generation of images from a textual prompt.
Heated debates surround its use in creating illustrations, photos, and comics. Some criticize “synographers” or “prompt engineers” for allegedly appropriating authors’ styles, while others defend the free enjoyment of culture.
Yearbook, Epik’s standout app filter, has dominated social feeds with a nostalgic wave of old yearbook photos. Along with Pixar’s Style, it’s part of the latest “deepfake” photo-editing craze, which allows users to create stunning visuals with their own faces.
At the forefront of this deepfake revolution is Aitana López, the AI-generated Spanish model and influencer who earns an impressive $10,000 a month. This trend has even infiltrated the realm of erotic content, pushing it into a perverse “singularity” with the automation of platforms like OnlyFans.
The collaboration between Runaway and the CompVis group at Ludwig Maximilians Universität in Munich has produced a standout model. Highly favored by digital production studios, artists, and syntographers, it now boasts a new feature - generating videos from a simple text prompt.
With open-source code, robust hardware, and the flexibility to train the model with custom datasets, creatives gain unparalleled control over their final output.
Revolutionizing film and video advertising this year, cutting-edge text-to-video and image-to-video software, led by Stable Diffusion’s ModelScope and Runaway 2, is reshaping the landscape.
Amidst the ironic and “cursed” videos of early pioneers surfacing on YouTube and Reddit, and Hollywood actors striking to safeguard their synthetic image rights, leading fashion and phone brands have engaged their creative departments in leveraging this groundbreaking technology.
Epic Games, the company behind Unreal Engine 5, the pinnacle of realistic graphics engines, have once again amazed us. This year’s surprise? Animator - a groundbreaking feature that allows users to create real-time animated facial models using only a smartphone camera. Animator, a powerful addition to MetaHuman Creator, the Unreal Engine 5-compatible software for creating lifelike human characters, is the result of Epic Games’ strategic acquisitions, including Quixel, home to the world’s largest photogrammetry repository.
Rockstar, the powerhouse behind the iconic Grand Theft Auto series, has just dropped a year-end bombshell on its fans: the GTA 6 trailer. Get ready for the stunning hyper-realism of RAGE 9, Rockstar’s graphics engine that is rumored to outshine even Unreal Engine 5 by jumping from 12k to a 16k resolution. Can it match the power of UE5’s game-changing Nanite? Only time will tell.
As of March 2023, OpenAI has introduced the upgraded ChatGPT (GPT-4) alongside the free version (GPT-3.5). This advanced model shows impressive capabilities, for example, when interacting with images and large text documents. Powered by multimodal applications such as integration with DALL-E 2, a “text-to-image” generator, this is just the beginning of a series of multitasking features that make the chatbot a frighteningly powerful tool.
The Internet of Things is teaming up with AI, transforming everything from smartwatches, robotic vacuums, and self-driving cars to voice assistants, digital wellbeing apps, medical devices, drones, and industrial robots. This evolution involves automating processes and integrating with AI software that adapts through user preferences or organizes into decentralized networks, known as swarm intelligence.