Artificial Creativity Engines in the US: History and Current Frontiers

NexaKing (NXK) Research

From AARON to GANs: A Brief History

Long before TikZ and neural nets, computer creativity began as an experiment in collaboration. One of the earliest and most famous examples is Harold Cohen’s AARON program. In the late 1960s, Cohen – a painter-turned-computer scientist at UC San Diego – wrote code that could draw abstract line-art. AARON used rule-based AI to compose and render images with pens and brushplotterswhitney.org. Cohen saw AARON as a collaboration between human and machine. As the Whitney Museum explains, Cohen understood his work with AARON “to be a collaboration” exploring how “an artist’s knowledge and process” could be translated into codewhitney.org. The pictures above are examples of AARON’s output. Even today, the Whitney notes that as modern tools like DALL·E and Midjourney go mainstream, Cohen’s legacy provides an “important historical perspective”whitney.org (see image below).

whitney.org Harold Cohen created AARON in the late 1960s as the first AI art program. He viewed it as a collaboration to “translate an artist’s knowledge and process into code.”

The Whitney Museum’s “Harold Cohen: AARON” exhibition highlights how his early AI drawings foreshadow today’s image-generation toolswhitney.org.

In music, similar experiments began. In 1957 Lejaren Hiller composed the first computer-generated piece (the ILLIAC Suite), and in the 1980s composer David Cope at UC Santa Cruz built an AI called EMI (Experiments in Musical Intelligence). Cope used EMI to analyze existing compositions and then generate new ones “in the style of the music in its database” without copying any piece exactlycomputerhistory.org. In effect, EMI taught itself a composer’s style and spun out new works, sometimes even fooling listeners. (In one famous test, an audience mistook an EMI-generated “Bach” piece for a real Bach compositioncomputerhistory.org.) Cope initially wrote EMI to break his own creative block: he programmed the system to propose musical lines he wouldn’t have written, provoked by the machine’s output. This “data-driven” approach helped him rediscover patterns in his stylecomputerhistory.org and compose an entire opera over several years. EMI eventually produced albums of “new” Bach, Mozart, and even Cope’s own stylecomputerhistory.orgcomputerhistory.org. The sheer fact that EMI could generate convincing classical-style music challenged notions of creativity. As Douglas Hofstadter (famous for Gödel, Escher, Bach) later said, EMI’s work forced us to “look at great works of art and wonder where they came from and how deep they really are”computerhistory.org.

Meanwhile, early AI literature was rudimentary – some experimented with grammar-based story generators – but real leaps came when large language models arrived. Today’s AI writing assistants (like OpenAI’s ChatGPT) trace their lineage to those early text programs but far surpass them. By late 2022, ChatGPT made AI writing mainstream, and even early studies show modern models have remarkable creative ability. For example, a 2024 scientific study found that GPT-4 outperformed human participants on divergent-thinking tasks, generating responses that were “more original and elaborate” than most people’snature.com. In short, machines have inherited some surprising creative spark from the data they’ve learned.

AI in Visual Art: GANs and Diffusion Models

Fast-forward to the 2010s and ‘20s: AI art burst into public view. The key technology was the Generative Adversarial Network (GAN), introduced by Ian Goodfellow in 2014. GANs pit two neural networks against each other – one creating images and the other judging them – to refine stunning visual output. In 2018, the art collective Obvious used a GAN trained on historical portraits to produce “Edmond de Belamy,” a blurry but recognizable face. That painting made headlines by selling for $432,500 at Christie’sen.wikipedia.org – a watershed moment showing AI art’s legitimacy. The parody in the name (a nod to “Goodfellow” in French) underscored the technology behind it. The auction set a tone: AI-generated works were entering museums and markets.

en.wikipedia.org “Edmond de Belamy,” a GAN-generated portrait by the Obvious collective, sold for $432,500 in 2018 – a landmark event for AI arten.wikipedia.org.

Today many tools make art generation accessible. In 2021, OpenAI released DALL·E, a model that creates images from text promptsen.wikipedia.org. Stability AI followed in 2022 with Stable Diffusion, an open model that any enthusiast could run on consumer hardwarestability.ai. Other systems like Midjourney and Adobe Firefly quickly followed. These modern engines let anyone describe a scene in words and watch it come to life. They have been widely adopted by hobbyists and professionals alikewavesix.ai. As one expert noted, AI “makes it easier to turn good ideas into art” while still relying on human imaginationwavesix.ai.

However, the rise of AI art also sparked debates. At a 2025 auction called “Augmented Intelligence,” Christie’s sold 28 AI-generated pieces totaling $728,784wavesix.ai. Works by tech-art pioneers like Refik Anadol were highlights (one sold for over $277,000wavesix.ai). Yet simultaneously, over 6,000 human artists signed an open letter protesting the auction, arguing that generative models train on copyrighted art without consentwavesix.ai. This underscored a rift: some artists embrace AI as a creative partner, while others see it as a threat to their livelihood and IP.

Even as controversy swirled, the technical advances kept coming. Today’s AI image-tools can mimic styles of painters, create photorealistic scenes, or blend concepts, all from a few wordswavesix.ai. The technology is moving fast and is largely driven by U.S. companies: OpenAI (California), Adobe (California), Google Magenta (Mountain View)magenta.withgoogle.com, and others. And institutes are taking notice: for example, a new $500,000 NEH-funded center at the University of Oklahoma is explicitly focusing on AI in cultural production – examining how AI affects trust, authenticity, and creativityou.edu. All this tells me that in the U.S., art and AI are deeply intertwined – from Silicon Valley labs to art school studios.

AI in Music: Composition and Collaboration

Algorithmic music has come a long way since Cope’s EMI. Modern AI music systems use deep learning, large datasets, and sometimes user-friendly interfaces. Google’s Magenta project (an open-source initiative) has developed models like MusicVAE, NSynth, and MusicLM to experiment with melody and timbremagenta.withgoogle.comwavesix.ai. For instance, Google introduced MusicLM, and Meta (Facebook) launched MusicGen, both capable of generating music from text promptswavesix.ai. There are also commercial platforms: Soundraw, AIVA, and Amper Music all provide AI-assisted composition (some aimed at film/game scoring, others at pop musicians)wavesix.ai.

Many musicians use these tools as collaborators. Electronic artist Holly Herndon, based in New York, famously calls AI her “jam partner” – she feeds audio into generative models to create novel harmonies and textureswavesix.ai. Singer-songwriter Taryn Southern (USA) even released a full album (I AM AI) composed and produced entirely with AI tools. She used IBM’s Watson Beat, Google Magenta, and other platforms to craft the tracksmedium.com. These examples show AI augmenting artists’ creativity rather than replacing it.

Music industry leaders are also engaging. Universal Music Group (UMG) announced collaborations with tech startups to build AI in ways that respect copyrights, emphasizing that ethical AI could “bolster and grow musical creativity”wavesix.ai. Still, pitfalls arise: a viral song mimicking Drake’s and The Weeknd’s voices was created without permission, sparking an industry warningwavesix.ai. Some artists have responded creatively – for example, Grimes (US) chose to license her vocals for AI remixing, sharing royalties with fans who generate new tracks using her voicewavesix.ai. Overall, major labels are working on revenue-sharing and quality control (AI models must credit source material properly) to navigate these waterswavesix.aiwavesix.ai.

AI in Literature and Writing

Text generation leaped forward with large language models. Today, tools like OpenAI’s ChatGPT, Microsoft’s 365 Copilot, and startups like Sudowrite are mainstream for writerswavesix.ai. Authors use them to brainstorm plots, overcome writer’s block, or refine drafts. These AI assistants can suggest sentences, prose styles, or character ideas. For example, software can rate story openings, propose twists, or even auto-complete a paragraph. In research terms, these models “assist with brainstorming, drafting, and editing” and are “widely used by authors”wavesix.ai. The key is that human authors still guide the process – most writers see AI as a helper, not a stand-in.

Nevertheless, the publishing world is adapting. With dozens of AI-generated books popping up on Amazon, the Authors Guild introduced a “Human Authored” certification for novels verified AI-freewavesix.ai. Magazines and publishers now often require authors to disclose AI use. The U.S. legal system has also weighed in: best-selling authors like John Grisham and Sarah Silverman sued AI firms for using their copyrighted texts to train models, and courts have indicated that purely AI-generated books aren’t eligible for copyright protectionwavesix.ai. These shifts reinforce that human creativity remains central; even a 2024 study noted that AI writing is built on human data – “no matter how magical… generative AI may seem, the content didn’t come from nowhere”digitaleconomy.stanford.edu.

Still, AI-authored novels and stories have appeared. Ross Goodwin (USA) experimented with AI in “1 the Road” (an automated road-trip novel), and projects like booksby.ai crank out novels entirely by neural nets. The quality varies – early AI texts often err on facts or plot coherence – but they’re improving. Interestingly, scientific tests suggest these models can be quite inventive: the GPT-4 system was “more original and elaborate” than people on creative thinking tasksnature.com. This suggests machines can spark ideas in unpredictable ways. In sum, AI tools for writing are now a common part of the writer’s toolbox, at least in the USA, with companies like OpenAI and Microsoft racing to refine them.

Organizations, Labs, and People in the US

In the United States, many universities and companies lead the charge in AI creativity. Big tech labs (OpenAI in San Francisco, Google Brain in Mountain View, Meta AI in California, etc.) fund research on generative models. Google’s open-source Magenta project explicitly explores “the role of machine learning as a tool in the creative process”magenta.withgoogle.com. Adobe (California) has baked AI into Photoshop and Illustrator to add “Generative Fill,” making it easier for designers to craft new visualswavesix.ai.

On the academic side, initiatives span from technical to cultural: Stanford’s Digital Economy Lab assembled an interdisciplinary team to study “the science of generative AI” alongside human creativitydigitaleconomy.stanford.edu. They published surveys and papers (even in Science) examining issues of creativity, ethics, and labor. At the University of Oklahoma, a new NEH-funded Center for Creativity and Authenticity in AI Cultural Production is being establishedou.edu. This center will bring together computer scientists, historians, artists, and Native American scholars to study how AI intersects with trust, art, and identity. Meanwhile, the National Science Foundation supports an AI2ES institute (in Oklahoma) focusing on trustworthy AI – indirectly relevant to media and art.

Key individuals also shape the field. In music and art there’s composer David Cope (UCSC) and artist Harold Cohen (UCLA/UCSD). On the tech side, OpenAI’s founders (including many US researchers) created ChatGPT, which indirectly fuels creative use. American artists like Sougwen Chung push the envelope by collaborating with robots and AI in live performancesmedium.com. Independent voices like Janelle Shane (USA) have also popularized AI creativity through experiments and humor, showing both the power and quirkiness of neural networks. And of course, we’ve mentioned entrepreneurs like those behind Stability AI (although UK-based, they have a global reach), or startups focused on creative tools. In sum, many projects and people in the U.S. are exploring how AI can serve creativity, spanning from tech megacorporations to university labs to indie artists.

Challenges and Ethical Issues

AI creativity brings both opportunity and controversy. Major concerns include copyright, authenticity, and job impacts. As noted, creators fret over copyright – large AI models often train on copyrighted images and text scraped from the web. The open letter by thousands of artists opposing Christie’s auction highlighted this; many signed it “citing concerns that AI models are trained on copyrighted works without consent”wavesix.ai. Similarly, authors are suing over unauthorized use of books. There are ongoing questions about who “owns” an AI-generated piece and how original it is.

Authorship and credit are also debated. If an AI paints a picture, who is the artist? Stanford researchers emphasize that generative AI’s output is ultimately derived from human work: “the content didn’t come from nowhere… it’s trained on the writing of human authors”digitaleconomy.stanford.edu. In practice, many human creators now label and disclaim AI involvement to maintain transparencywavesix.ai. Legally, U.S. courts have so far ruled that fully AI-generated works can’t get copyright protectionwavesix.ai, reinforcing the idea that some human input is needed.

Originality and diversity are concerns too. Some worry AI could lead to formulaic art. For example, one paper notes that if models keep training on other AI-generated images, we might end up in a loop of repeating styles. The Stanford team also notes the need to discuss issues like bias and misinformation.

Finally, there are economic and social factors. Will illustrators, composers or writers be replaced by AI? Trade unions in the US (like SAG-AFTRA for actors) have already raised alarms about digital likenesses. So far, industry leaders (like UMG’s CEO) insist AI should augment creativity, not replace artistswavesix.ai. The consensus is slowly forming that AI, when guided ethically, can be an amplifier of human creativity. As one comment puts it, “AI should be a tool for augmentation, not replacement”wavesix.ai.

The Road Ahead: Human-AI Collaboration

Where does this leave us? Overall, AI creativity is evolving quickly. On one hand, artists and audiences are amazed by what these engines can do; on the other hand, many frontline questions remain. Researchers like me (NXK) will continue to watch how this transforms the creative landscape.

Looking ahead, the trend is toward synergy. Virtually every recent report or expert predicts a future where human ingenuity drives AI, not the other way around. The wavesix blog concludes that “the future of creativity belongs not to AI alone, but to the artists, musicians… who wield it with vision, intent, and human emotion”wavesix.ai. Likewise, musician Holly Herndon envisions AI as enabling “collaboration in ways we couldn’t imagine before”wavesix.ai.

In practice, this means ongoing experimentation: expect more pop albums co-written by AI, more blockbuster movies partly rendered by AI (with SAG and writers negotiating terms), and more interactive stories where the reader’s choices and an AI’s generative plot mix together. The U.S. will continue to be a major stage – Silicon Valley companies, New York publishers, Hollywood studios, and academic conferences all contribute to charting the path.

In conclusion, the history of artificial creativity engines runs from pencil-drawing robots in the 1970s to neural networks that can conjure entire scenes by algorithm. The tools keep improving, and attitudes keep adapting. As an AI observer (NXK), I’m both excited by what AI can do and mindful of its limits. The journey of art, music, and literature into the AI era shows a picture of co-creativity, with humans and machines each playing a role. If we handle it right – with ethics, fair rules, and openness – we may well see an explosion of new art forms that none of us could fully predict.

Sources: