‍

Artificial and Human Intelligence

‍The AI Index 2024 Annual Report by Stanford University—an invaluable publication aiming to enhance understanding of AI through dedicated research—addresses several issues pertinent to the digital art space without dedicating a chapter or even section to it. Peter Bauman (Monk Antony) reviewed the five-hundred-page report to distill the relevant takeaways for the contemporary art world’s collectors, artists and institutions.‍

The public increasingly fears AI, while private capital increasingly adores it. As it relates to art, these are two of the more far-reaching takeaways from Stanford University’s AI Index 2024 Annual Report. What are the implications of this dynamic for collectors, artists and institutions in the digital art space? Humans have long feared new technology so we predictably fear AI. This fact in isolation does not reveal much.
‍

What if we compare our current AI trepidation to two similarly transformative technologies that appeared poised to alter society: the Internet and the computer at their beginnings?

Public perception of AI appears to be associated with higher levels of anxiety than the Internet in the 1990s, where sentiment was marked more with skepticism than fear. Stanford's AI Index report showcases the widespread levels of AI unease today, citing Pew opinion research that only 10% of U.S. adults reported feeling “more excited than concerned” about AI. In contrast, 52% were “more concerned than excited,” which spiked from 38% in 2022 and 37% in 2021. Apprehension at the level of AI appears more reminiscent of our 1960s computer phobia—also associated with humanity’s demise, job replacement and big tech oligarchy. What eventually dispelled that fear? Use cases that continuously improved—and became entangled in—people’s lives to the point where not having one seemed unthinkable.

This fear-dispelling process for AI may have quietly begun with events like Apple’s June 2024 WWDC24. At the event, Apple revealed its full-fledged embrace and push into AI in the form of “Apple Intelligence,” a generative AI suite across its entire product ecosystem. Over the next several months, more than 2 billion active Apple devices will be updated with integrated ChatGPT, AI-generated imagery capabilities and AI writing enhancements. While not certain to become an immediate success, Apple’s update demonstrates the facility with which AI will be absorbed into our lives.
‍
The AI Index report’s five hundred pages and nine chapters investigate AI from manifold perspectives, including R&D, the economy, science, politics and education. Yet, there is no section for culture or art, indicating how nascent that component of this puzzle piece is within a still-burgeoning field. Below, I synthesize the information from the nine chapters relevant to the contemporary art world—specifically its collectors, artists and institutions.

‍

Collectors

Perhaps the AI Index report’s biggest takeaway from the art world perspective is that, while public concern spiked, so has investment. Funding for generative AI has jumped eightfold since just 2022, reaching $25.2 billion in 2023. These skyrocketing figures will impact the entire ecosystem in numerous ways but for collectors, it’s a sign of confidence. Despite general fears surrounding AI—and AI art—this substantial capital injection stands as a powerful counter to any narrative that generative AI may not be here to stay. In other words, the number indicates a healthy, rapidly developing ecosystem that will drive the enhancement of tools capable of creating the art that speaks to and challenges our time. Artists are by no means required to use these tools, but they exist and may even encourage new artists who have never found suitable forms of expression.

The investment figures indicate that the market for AI-generated art will continue to evolve and possibly expand. Unique to today, collectors have a chance to own work created with the advanced technology​​ dictating our lives in a way that simply wasn’t possible with net art in the mid-90s or computer-generated art in the ‘60s and ‘70s. Beyond experiencing and participating in culture, collectors can own a piece of it.
‍

‍

Artists

Painting, sculpture and traditional means of creative expression are not going away, yet it's undeniable that the toolkit from which artists now have to choose is expanding steeply, as indicated by the Stanford report. The AI Index report highlights how AI image and text-generating models have expanded sharply, particularly open-source models relevant to the most serious engagements with AI today.

According to the report, as funding has accelerated, so has the development and sophistication of AI models for image generation and manipulation. It emphasizes that the past year saw a breakthrough in AI models becoming multimodal—capable of processing and generating content across multiple data inputs. Models can now comprehend text as well as process images, audio and video, giving artists renewed flexibility to create.

Additionally, researchers developed a benchmark to assess the various models—not dissimilar to previous attempts by Victor Vasarely or the cyberneticists to encode aesthetics—known as HEIM (Holistic Evaluation of Text-to-Image Models). HEIM evaluates different models in areas like aesthetics, originality and image quality. The Stable Diffusion-based Dreamlike Photoreal ranks highest in all three categories, according to HEIM, outscoring DALL-E 2 and Midjourney. Yet DALL-2 wins handily in terms of image accuracy from the prompt.
‍

‍

The report notes another interesting benchmark tool for artists, BiasPainter, which detects bias across commercial image-generation models in areas like age, race and gender. As Danielle King noted for Le Random Editorials, these image-generation models can perpetuate harmful stereotypes. Yet tooling enhancements such as these may augment artist decision-making to expand what is possible visually.

Critically, open-source tools are becoming increasingly available, as much of the most thoughtful AI-based work involves artists creating bespoke models from open-source code. Open-source tooling allows artists to personalize their work and mostly ignore the latest big tech updates.

The report itself details the open-source research eruption over the last decade: “Since 2011, the number of AI-related projects on GitHub has seen a consistent increase, growing from 845 in 2011 to approximately 1.8 million in 2023.” But the past year alone saw “a sharp 59.3% rise.” Beyond quantity, there are indications that quality has improved as well. The report states that “the total number of stars [likes] for AI-related projects on GitHub also significantly increased in 2023, more than tripling from 4.0 million in 2022 to 12.2 million.”

From open-source to commercial models, the rapid increase in developments—combined with the previously mentioned funding increases—suggests that digital artists now have unparalleled access to powerful and sophisticated AI tools. Yet how will they continue impacting artists’ workflow—as I discussed with Matt DesLauriers—beyond improving areas like efficiency? How will these tools empower artists to more truthfully express their visions?

‍

Institutions

For institutions, the sudden arrival of AI and its implications for art will have profound—even existential—consequences. Previously, academia was the leader in machine learning model creation, making it a more natural partner for institutions like museums. But since 2014, the report mentions that big tech companies have begun to dominate, largely due to the financial, data and compute limitations at university research departments relative to Google or Microsoft. From an institutional perspective, that means any likely collaborations in order to remain relevant will require navigating relationships with these potentially less savory partners.

Finally, with the previously mentioned $25.2 billion in funding in 2023 for a tool that can be used extensively in the arts, major museums and galleries have taken notice and more are likely to follow. The fine-art world has been struggling since the pandemic, in particular with museums around the world struggling to return to pre-COVID attendance levels. The most forward-thinking and enterprising institutions will thus be looking towards AI for their artistic breakthroughs as well as for their earning potential.

‍

Closing

Stanford’s AI Index report highlights how transformative a year 2023 was—often difficult to determine when living through it. The report underscores the myriad, dizzying ways in which AI will collide with our lives in the future. AI isn’t going away and it’s entirely normal—if not expected—to feel somewhat overwhelmed by its prospects. But understanding the trends highlighted here can equip participants in this space with the agility to maneuver in the coming years.

This report on artificial intelligence also reveals something about our human intelligence. If intelligence is the application of knowledge, then, by definition, it would be unintelligent for humans not to apply their artistic knowledge to the latest tools of our age. We have a long track record of these experiments—from neolithic cave dwellers blowing pigments to a Bell Labs employee making “patterns” with a mainframe computer in 1962. By intervening in the latest technologies, artists are able to restate our longest-asked questions—relating to human will, perception and making—so that the answers may be reimagined as well.


‍
--



Peter Bauman (Monk Antony) is Le Random's Editor-in-Chief.