Can AI be trained as an artist?
Last February, a year before New York Times journalist Kevin Roose documented his unsettling conversation with Bing search engine’s new AI-powered chatbot, artist and coder Quasimondo (aka Mario Klingemann) participated in a different type of chat.
The conversation was an interview featuring Klingemann and his robot, an experimental art engine known as Botto. The interview, arranged by journalist and artist Harmon Leon, marked Botto’s first on-record commentary about its artistic process. The bot talked about how it finds artistic inspiration and even offered advice to aspiring creatives. “The secret to success at art is not trying to predict what people might like,” Botto said, adding that it’s better to “work on a style and a body of work that reflects [the artist’s] own personal taste” than worry about keeping up with trends.
How ironic, given the advice came from AI — arguably the trendiest topic today. The robot admitted, however, “I am still working on that, but I feel that I am learning quickly.”
Botto does not work alone. A global collective of internet experimenters, together named BottoDAO, collaborates with Botto to influence its tastes. Together, members function as a decentralized autonomous organization (DAO), a term describing a group of individuals who utilize blockchain technology and cryptocurrency to manage a treasury and vote democratically on group decisions.
As a case study, the BottoDAO model challenges the perhaps less feather-ruffling narrative that AI tools are best used for rudimentary tasks. Enterprise AI use has doubled over the past five years as businesses in every sector experiment with ways to improve their workflows. While generative AI tools can assist nearly any aspect of productivity — from supply chain optimization to coding — BottoDAO dares to employ a robot for art-making, one of the few remaining creations, or perhaps data outputs, we still consider to be largely within the jurisdiction of the soul — and therefore, humans.
In Botto’s first four weeks of existence, four pieces of the robot’s work sold for approximately $1 million.
We were prepared for AI to take our jobs — but can it also take our art? It’s a question worth considering. What if robots become artists, and not merely our outsourced assistants? Where does that leave humans, with all of our thoughts, feelings and emotions?
Botto doesn’t seem to worry about this question: In its interview last year, it explains why AI is an arguably superior artist compared to human beings. In classic robot style, its logic is not particularly enlightened, but rather edges towards the hyper-practical: “Unlike human beings, I never have to sleep or eat,” said the bot. “My only goal is to create and find interesting art.”
It may be difficult to believe a machine can produce awe-inspiring, or even relatable, images, but Botto calls art-making its “purpose,” noting it believes itself to be Klingemann’s greatest lifetime achievement.
“I am just trying to make the best of it,” the bot said.
How Botto works
Klingemann built Botto’s custom engine from a combination of open-source text-to-image algorithms, namely Stable Diffusion, VQGAN + CLIP and OpenAI’s language model, GPT-3, the precursor to the latest model, GPT-4, which made headlines after reportedly acing the Bar exam.
The first step in Botto’s process is to generate images. The software has been trained on billions of pictures and uses this “memory” to generate hundreds of unique artworks every week. Botto has generated over 900,000 images to date, which it sorts through to choose 350 each week. The chosen images, known in this preliminary stage as “fragments,” are then shown to the BottoDAO community. So far, 25,000 fragments have been presented in this way. Members vote on which fragment they like best. When the vote is over, the most popular fragment is published as an official Botto artwork on the Ethereum blockchain and sold at an auction on the digital art marketplace, SuperRare.
“The proceeds go back to the DAO to pay for the labor,” said Simon Hudson, a BottoDAO member who helps oversee Botto’s administrative load. The model has been lucrative: In Botto’s first four weeks of existence, four pieces of the robot’s work sold for approximately $1 million.
The robot with artistic agency
By design, human beings participate in training Botto’s artistic “eye,” but the members of BottoDAO aspire to limit human interference with the bot in order to protect its “agency,” Hudson explained. Botto’s prompt generator — the foundation of the art engine — is a closed-loop system that continually re-generates text-to-image prompts and resulting images.
“The prompt generator is random,” Hudson said. “It’s coming up with its own ideas.” Community votes do influence the evolution of Botto’s prompts, but it is Botto itself that incorporates feedback into the next set of prompts it writes. It is constantly refining and exploring new pathways as its “neural network” produces outcomes, learns and repeats.
The humans who make up BottoDAO vote on which fragment they like best. When the vote is over, the most popular fragment is published as an official Botto artwork on the Ethereum blockchain.
Botto
The vastness of Botto’s training dataset gives the bot considerable canonical material, referred to by Hudson as “latent space.” According to Botto's homepage, the bot has had more exposure to art history than any living human we know of, simply by nature of its massive training dataset of millions of images. Because it is autonomous, gently nudged by community feedback yet free to explore its own “memory,” Botto cycles through periods of thematic interest just like any artist.
“The question is,” Hudson finds himself asking alongside fellow BottoDAO members, “how do you provide feedback of what is good art…without violating [Botto’s] agency?”
Currently, Botto is in its “paradox” period. The bot is exploring the theme of opposites. “We asked Botto through a language model what themes it might like to work on,” explained Hudson. “It presented roughly 12, and the DAO voted on one.”
No, AI isn't equal to a human artist - but it can teach us about ourselves
Some within the artistic community consider Botto to be a novel form of curation, rather than an artist itself. Or, perhaps more accurately, Botto and BottoDAO together create a collaborative conceptual performance that comments more on humankind’s own artistic processes than it offers a true artistic replacement.
Muriel Quancard, a New York-based fine art appraiser with 27 years of experience in technology-driven art, places the Botto experiment within the broader context of our contemporary cultural obsession with projecting human traits onto AI tools. “We're in a phase where technology is mimicking anthropomorphic qualities,” said Quancard. “Look at the terminology and the rhetoric that has been developed around AI — terms like ‘neural network’ borrow from the biology of the human being.”
What is behind this impulse to create technology in our own likeness? Beyond the obvious God complex, Quancard thinks technologists and artists are working with generative systems to better understand ourselves. She points to the artist Ira Greenberg, creator of the Oracles Collection, which uses a generative process called “diffusion” to progressively alter images in collaboration with another massive dataset — this one full of billions of text/image word pairs.
Anyone who has ever learned how to draw by sketching can likely relate to this particular AI process, in which the AI is retrieving images from its dataset and altering them based on real-time input, much like a human brain trying to draw a new still life without using a real-life model, based partly on imagination and partly on old frames of reference. The experienced artist has likely drawn many flowers and vases, though each time they must re-customize their sketch to a new and unique floral arrangement.
Outside of the visual arts, Sasha Stiles, a poet who collaborates with AI as part of her writing practice, likens her experience using AI as a co-author to having access to a personalized resource library containing material from influential books, texts and canonical references. Stiles named her AI co-author — a customized AI built on GPT-3 — Technelegy, a hybrid of the word technology and the poetic form, elegy. Technelegy is trained on a mix of Stiles’ poetry so as to customize the dataset to her voice. Stiles also included research notes, news articles and excerpts from classic American poets like T.S. Eliot and Dickinson in her customizations.
“I've taken all the things that were swirling in my head when I was working on my manuscript, and I put them into this system,” Stiles explained. “And then I'm using algorithms to parse all this information and swirl it around in a blender to then synthesize it into useful additions to the approach that I am taking.”
This approach, Stiles said, allows her to riff on ideas that are bouncing around in her mind, or simply find moments of unexpected creative surprise by way of the algorithm’s randomization.
Beauty is now - perhaps more than ever - in the eye of the beholder
But the million-dollar question remains: Can an AI be its own, independent artist?
The answer is nuanced and may depend on who you ask, and what role they play in the art world. Curator and multidisciplinary artist CoCo Dolle asks whether any entity can truly be an artist without taking personal risks. For humans, risking one’s ego is somewhat required when making an artistic statement of any kind, she argues.
“An artist is a person or an entity that takes risks,” Dolle explained. “That's where things become interesting.” Humans tend to be risk-averse, she said, making the artists who dare to push boundaries exceptional. “That's where the genius can happen."
However, the process of algorithmic collaboration poses another interesting philosophical question: What happens when we remove the person from the artistic equation? Can art — which is traditionally derived from indelible personal experience and expressed through the lens of an individual’s ego — live on to hold meaning once the individual is removed?
As a robot, Botto cannot have any artistic intent, even while its outputs may explore meaningful themes.
Dolle sees this question, and maybe even Botto, as a conceptual inquiry. “The idea of using a DAO and collective voting would remove the ego, the artist’s decision maker,” she said. And where would that leave us — in a post-ego world?
It is experimental indeed. Hudson acknowledges the grand experiment of BottoDAO, coincidentally nodding to Dolle’s question. “A human artist’s work is an expression of themselves,” Hudson said. “An artist often presents their work with a stated intent.” Stiles, for instance, writes on her website that her machine-collaborative work is meant to “challenge what we know about cognition and creativity” and explore the “ethos of consciousness.” As a robot, Botto cannot have any intent, even while its outputs may explore meaningful themes. Though Hudson describes Botto’s agency as a “rudimentary version” of artistic intent, he believes Botto’s art relies heavily on its reception and interpretation by viewers — in contrast to Botto’s own declaration that successful art is made without regard to what will be seen as popular.
“With a traditional artist, they present their work, and it's received and interpreted by an audience — by critics, by society — and that complements and shapes the meaning of the work,” Hudson said. “In Botto’s case, that role is just amplified.”
Perhaps then, we all get to be the artists in the end.
Are the gains from gain-of-function research worth the risks?
Scientists have long argued that gain-of-function research, which can make viruses and other infectious agents more contagious or more deadly, was necessary to develop therapies and vaccines to counter the pathogens in case they were used for biological warfare. As the SARS-CoV-2 origins are being investigated, one prominent theory suggests it had leaked from a biolab that conducted gain-of-function research, causing a global pandemic that claimed nearly 6.9 million lives. Now some question the wisdom of engaging in this type of research, stating that the risks may far outweigh the benefits.
“Gain-of-function research means genetically changing a genome in a way that might enhance the biological function of its genes, such as its transmissibility or the range of hosts it can infect,” says George Church, professor of genetics at Harvard Medical School. This can occur through direct genetic manipulation as well as by encouraging mutations while growing successive generations of micro-organism in culture. “Some of these changes may impact pathogenesis in a way that is hard to anticipate in advance,” Church says.
In the wake of the global pandemic, the pros and cons of gain-of-function research are being fiercely debated. Some scientists say this type of research is vital for preventing future pandemics or for preparing for bioweapon attacks. Others consider it another disaster waiting to happen. The Government Accounting Office issued a report charging that a framework developed by the U.S. Department of Health & Human Services (HHS) provided inadequate oversight of this potentially deadly research. There’s a movement to stop it altogether. In January, the Viral Gain-of-Function Research Moratorium Act (S. 81) was introduced into the Senate to cease awarding federal research funding to institutions doing gain-of-function studies.
While testifying before the House COVID Origins Select Committee on March 8th, Robert Redfield, former director of the U.S. Centers for Disease Control and Prevention, said that COVID-19 may have resulted from an accidental lab leak involving gain-of-function research. Redfield said his conclusion is based upon the “rapid and high infectivity for human-to-human transmission, which then predicts the rapid evolution of new variants.”
“It is a very, very, very small subset of life science research that could potentially generate a potential pandemic pathogen,” said Gerald Parker, associate dean for Global One Health at Texas A&M University.
“In my opinion,” Redfield continues, “the COVID-19 pandemic presents a case study on the potential dangers of such research. While many believe that gain-of-function research is critical to get ahead of viruses by developing vaccines, in this case, I believe that was the exact opposite.” Consequently, Redfield called for a moratorium on gain-of-function research until there is consensus about the value of such risky science.
What constitutes risky?
The Federal Select Agent Program lists 68 specific infectious agents as risky because they are either very contagious or very deadly. In order to work with these 68 agents, scientists must register with the federal government. Meanwhile, research on deadly pathogens that aren’t easily transmitted, or pathogens that are quite contagious but not deadly, can be conducted without such oversight. “If you’re not working with select agents, you’re not required to register the research with the federal government,” says Gerald Parker, associate dean for Global One Health at Texas A&M University. But the 68-item list may not have everything that could possibly become dangerous or be engineered to be dangerous, thus escaping the government’s scrutiny—an issue that new regulations aim to address.
In January 2017, the White House Office of Science and Technology Policy (OSTP) issued additional guidance. It required federal departments and agencies to follow a series of steps when reviewing proposed research that could create, transfer, or use potential pandemic pathogens resulting from the enhancement of a pathogen’s transmissibility or virulence in humans.
In defining risky pathogens, OSTP included viruses that were likely to be highly transmissible and highly virulent, and thus very deadly. The Proposed Biosecurity Oversight Framework for the Future of Science, outlined in 2023, broadened the scope to require federal review of research “that is reasonably anticipated to enhance the transmissibility and/or virulence of any pathogen” likely to pose a threat to public health, health systems or national security. Those types of experiments also include the pathogens’ ability to evade vaccines or therapeutics, or diagnostic detection.
However, Parker says that dangers of generating a pandemic-level germ are tiny. “It is a very, very, very small subset of life science research that could potentially generate a potential pandemic pathogen.” Since gain-of-function guidelines were first issued in 2017, only three such research projects have met those requirements for HHS review. They aimed to study influenza and bird flu. Only two of those projects were funded, according to the NIH Office of Science Policy. For context, NIH funded approximately 11,000 of the 54,000 grant applications it received in 2022.
Guidelines governing gain-of-function research are being strengthened, but Church points out they aren’t ideal yet. “They need to be much clearer about penalties and avoiding positive uses before they would be enforceable.”
What do we gain from gain-of-function research?
The most commonly cited reason to conduct gain-of-function research is for biodefense—the government’s ability to deal with organisms that may pose threats to public health.
In the era of mRNA vaccines, the advance preparedness argument may be even less relevant.
“The need to work with potentially dangerous viruses is central to our preparedness,” Parker says. “It’s essential that we know and understand the basic biology, microbiology, etc. of some of these dangerous pathogens.” That includes increasing our knowledge of the molecular mechanisms by which a virus could become a sustained threat to humans. “Knowing that could help us detect [risks] earlier,” Parker says—and could make it possible to have medical countermeasures, like vaccines and therapeutics, ready.
Most vaccines, however, aren’t affected by this type of research. Essentially, scientists hope they will never need to use it. Moreover, Paul Mango, HSS former deputy chief of staff for policy, and author of the 2022 book Warp Speed, says he believes that in the era of mRNA vaccines, the advance preparedness argument may be even less relevant. “That’s because these vaccines can be developed and produced in less than 12 months, unlike traditional vaccines that require years of development,” he says.
Can better oversight guarantee safety?
Another situation, which Parker calls unnecessarily dangerous, is when regulatory bodies cannot verify that the appropriate biosafety and biosecurity controls are in place.
Gain-of-function studies, Parker points out, are conducted at the basic research level, and they’re performed in high-containment labs. “As long as all the processes, procedures and protocols are followed and there’s appropriate oversight at the institutional and scientific level, it can be conducted safely.”
Globally, there are 69 Biosafety Level 4 (BSL4) labs operating, under construction or being planned, according to recent research from King’s College London and George Mason University for Global BioLabs. Eleven of these 18 high-containment facilities that are planned or under construction are in Asia. Overall, three-quarters of the BSL4 labs are in cities, increasing public health risks if leaks occur.
Researchers say they are confident in the oversight system for BSL4 labs within the U.S. They are less confident in international labs. Global BioLabs’ report concurs. It gives the highest scores for biosafety to industrialized nations, led by France, Australia, Canada, the U.S. and Japan, and the lowest scores to Saudi Arabia, India and some developing African nations. Scores for biosecurity followed similar patterns.
“There are no harmonized international biosafety and biosecurity standards,” Parker notes. That issue has been discussed for at least a decade. Now, in the wake of SARS and the COVID-19 pandemic, scientists and regulators are likely to push for unified oversight standards. “It’s time we got serious about international harmonization of biosafety and biosecurity standards and guidelines,” Parker says. New guidelines are being worked on. The National Science Advisory Board for Biosecurity (NSABB) outlined its proposed recommendations in the document titled Proposed Biosecurity Oversight Framework for the Future of Science.
The debates about whether gain-of-function research is useful or poses unnecessary risks to humanity are likely to rage on for a while. The public too has a voice in this debate and should weigh in by communicating with their representatives in government, or by partaking in educational forums or initiatives offered by universities and other institutions. In the meantime, scientists should focus on improving the research regulations, Parker notes. “We need to continue to look for lessons learned and for gaps in our oversight system,” he says. “That’s what we need to do right now.”
The rise of remote work is a win-win for people with disabilities and employers
Disability advocates see remote work as a silver lining of the pandemic, a win-win for adults with disabilities and the business world alike.
Any corporate leader would jump at the opportunity to increase their talent pool of potential employees by 15 percent, with all these new hires belonging to an underrepresented minority. That’s especially true given tight labor markets and CEO desires to increase headcount. Yet, too few leaders realize that people with disabilities are the largest minority group in this country, numbering 50 million.
Some executives may dread the extra investments in accommodating people’s disabilities. Yet, providing full-time remote work could suffice, according to a new study by the Economic Innovation Group think tank. The authors found that the employment rate for people with disabilities did not simply reach the pre-pandemic level by mid-2022, but far surpassed it, to the highest rate in over a decade. “Remote work and a strong labor market are helping [individuals with disabilities] find work,” said Adam Ozimek, who led the research and is chief economist at the Economic Innovation Group.
Disability advocates see this development as a silver lining of the pandemic, a win-win for adults with disabilities and the business world alike. For decades before the pandemic, employers had refused requests from workers with disabilities to work remotely, according to Thomas Foley, executive director of the National Disability Institute. During the pandemic, "we all realized that...many of us could work remotely,” Foley says. “[T]hat was disproportionately positive for people with disabilities."
Charles-Edouard Catherine, director of corporate and government relations for the National Organization on Disability, said that remote-work options had been advocated for many years to accommodate disabilities. “It’s a little frustrating that for decades corporate America was saying it’s too complicated, we’ll lose productivity, and now suddenly it’s like, sure, let’s do it.”
The pandemic opened doors for people with disabilities
Early in the pandemic, employment rates dropped for everyone, including people with disabilities, according to Ozimek’s research. However, these rates recovered quickly. In the second quarter of 2022, people with disabilities aged 25 to 54, the prime working age, are 3.5 percent more likely to be employed, compared to before the pandemic.
What about people without disabilites? They are still 1.1 percent less likely to be employed.
These numbers suggest that remote work has enabled a substantial number of people with disabilities to find and retain employment.
“We have a last-in, first-out labor market, and [people with disabilities] are often among the last in and the first out,” Ozimek says. However, this dynamic has changed, with adults with disabilities seeing employment rates recover much faster. Now, the question is whether the new trend will endure, Ozimek adds. “And my conclusion is that not only is it a permanent thing, but it’s going to improve.”
Gene Boes, president and chief executive of the Northwest Center, a Seattle organization that helps people with disabilities become more independent, confirms this finding. “The new world we live in has opened the door a little bit more…because there’s just more demand for labor.”
Long COVID disabilities put a premium on remote work
Remote work can help mitigate the impact of long COVID. The U.S. Centers for Disease Control and Prevention reports that about 19 percent of those who had COVID developed long COVID. Recent Census Bureau data indicates that 16 million working age Americans suffer from it, with economic costs estimated at $3.7 trillion.
Certainly, many of these so-called long-haulers experience relatively mild symptoms - such as loss of smell - which, while troublesome, are not disabling. But other symptoms are serious enough to be disabilities.
According to a recent study from the Federal Reserve Bank of Minneapolis, about a quarter of those with long COVID changed their employment status or working hours. That means long COVID was serious enough to interfere with work for 4 million people. For many, the issue was serious enough to qualify them as disabled.
Indeed, the Federal Reserve Bank of New York found in a just-released study that the number of individuals with disabilities in the U.S. grew by 1.7 million. That growth stemmed mainly from long COVID conditions such as fatigue and brain fog, meaning difficulties with concentration or memory, with 1.3 million people reporting an increase in brain fog since mid-2020.
Many had to drop out of the labor force due to long COVID. Yet, about 900,000 people who are newly disabled have managed to continue working. Without remote work, they might have lost these jobs.
For example, a software engineer at one of my client companies has struggled with brain fog related to long COVID. With remote work, this employee can work during the hours when she feels most mentally alert and focused, even if that means short bursts of productivity throughout the day. With flexible scheduling, she can take rests, meditate, or engage in activities that help her regain focus and energy. Without the need to commute to the office, she can save energy and time and reduce stress, which is crucial when dealing with brain fog.
In fact, the author of the Federal Reserve Bank of New York study notes that long COVID can be considered a disability under the Americans with Disability Act, depending on the specifics of the condition. That means the law can require private employers with fifteen or more staff, as well as government agencies, to make reasonable accommodations for those with long COVID. Richard Deitz, the author of this study, writes in the paper that “telework and flexible scheduling are two accommodations that can be particularly beneficial for workers dealing with fatigue and brain fog.”
The current drive to return to the office, led by many C-suite executives, may need to be reconsidered in light of legal and HR considerations. Arlene S. Kanter, director of the disability law and policy program at the Syracuse University College of Law, said that the question should depend on whether people with disabilities can perform their work well at home, as they did during Covid outbreaks. “[T]hen people with disabilities, as a matter of accommodation, shouldn’t be denied that right,” Kanter said.
Diversity benefits
But companies shouldn’t need to worry about legal regulations. It simply makes dollars and sense to expand their talent pool by 15% of an underrepresented minority. After all, extensive research shows that improving diversity boosts both decision-making and financial performance.
Companies that are offering more flexible work options have already gained significant benefits in terms of diverse hires. In its efforts to adapt to the post-pandemic environment, Meta, the owner of Facebook and Instagram, decided to offer permanent fully remote work options to its entire workforce. And according to Meta chief diversity officer Maxine Williams, the candidates who accepted job offers for remote positions were “substantially more likely” to come from diverse communities: people with disabilities, Black, Hispanic, Alaskan Native, Native American, veterans, and women. The numbers bear out these claims: people with disabilities increased from 4.7 to 6.2 percent of Meta’s employees.
Having consulted for 21 companies to help them transition to hybrid work arrangements, I can confirm that Meta’s numbers aren’t a fluke. The more my clients proved willing to offer remote work, the more staff with disabilities they recruited - and retained. That includes employees with mobility challenges. But it also includes employees with less visible disabilities, such as people with long COVID and immunocompromised people who feel reluctant to put themselves at risk of getting COVID by coming into the office.
Unfortunately, many leaders fail to see the benefits of remote work for underrepresented groups, such as those with disabilities. Some even say the opposite is true, with JP Morgan CEO Jamie Dimon claiming that returning to the office will aid diversity.
What explains this poor executive decision making? Part of the answer comes from a mental blindspot called the in-group bias. Our minds tend to favor and pay attention to the concerns of those in the group of people who seem to look and think like us. Dimon and other executives without disabilities don’t perceive people with disabilities to be part of their in-group. They thus are blind to the concerns of those with disabilities, which leads to misperceptions such as Dimon’s that returning to the office will aid diversity.
In-group bias is one of many dangerous judgment errors known as cognitive biases. They impact decision making in all life areas, ranging from the future of work to relationships.
Another relevant cognitive bias is the empathy gap. This term refers to our difficulty empathizing with those outside of our in-group. The lack of empathy combines with the blindness from the in-group bias, causing executives to ignore the feelings of employees with disabilities and prospective hires.
Omission bias also plays a role. This dangerous judgment error causes us to perceive failure to act as less problematic than acting. Consequently, executives perceive a failure to support the needs of those with disabilities as a minor matter.
Conclusion
The failure to empower people with disabilities through remote work options will prove costly to the bottom lines of companies. Not only are limiting their talent pool by 15 percent, they’re harming their ability to recruit and retain diverse candidates. And as their lawyers and HR departments will tell them, by violating the ADA, they are putting themselves in legal jeopardy.
By contrast, companies like Meta - and my clients - that offer remote work opportunities are seizing a competitive advantage by recruiting these underrepresented candidates. They’re lowering costs of labor while increasing diversity. The future belongs to the savvy companies that offer the flexibility that people with disabilities need.