Story—sacred and profane—is perhaps the main cohering force in human life. A society is composed of fractious people with different personalities, goals, and agendas. What connects us beyond our kinship ties? Story. … Story is the counterforce to social disorder, the tendency of things to fall apart. Story is the center without which the rest cannot hold.
— Jonathan Gottschall, The Storytelling Animal: How Stories Make Us Human
Mythology, in other words, is psychology misread as biography, history, and cosmology.
— Joseph Campbell, The Hero with a Thousand Faces
This is the introduction to a much longer article on the legal consequences of generative AI which can be found here.
The question was simple: how does one explain to a 9-year-old what the James Webb Space Telescope has discovered? The answer should have been as well. But Google Bard got it wrong, and the embarrassment cost the company $100 billion in market value.[1] That is the indelible lesson from the Google preview of Google Bard, the generative artificial intelligence (AI) program designed to use natural language to glean facts from across the internet and provide interactive chat with users on whatever subjects might cross their minds.
Bard incorrectly stated that the James Webb telescope took the “very first pictures of a planet outside of our own solar system.”[2] The problem with the answer is that the Very Large Telescope in Chile actually earned this scientific first.[3] The answer shook a stock market that had been betting on Alphabet having a technological answer ready to compete with OpenAI and the incorporation of ChatGPT technology into Bing and Office.
The stock market had been giddy with the prospect of Microsoft, Alphabet, Meta, and Nvidia (maker of the graphical processing units providing most of the hardware for next generation AI)[4] and the hundreds of companies trying to participate in the cultural phenomenon unleashed by the public launch of OpenAI’s ChatGPT.[5] “ChatGPT’s wide range of uses, status as the fastest-growing consumer app in history, and potential to disrupt internet searches has pushed rivals to throw out the slow and cautious strategy that has dominated A.I. research for years.”[6]
Using a variety of different training methods (which are discussed in section 3), new and emerging generative AI systems can produce data that mimics the content of human creativity.[7] These neural networks can generate content in the form of text, voice, pictures, videos, software, physical and molecular designs, audiovisual works combining these features, and more.[8]
As a story-telling medium, “ChatGPT has been used to create thousands of books that have been published on Amazon. Interestingly enough, the genres and topics covered by these AI-generated books are quite wide, ranging from self-help to cookbooks to romance novels.”[9] The services are far more than search engines because they do not merely find published content, they evaluate, combine, and synthesize the known information to provide an answer of their own. Trained on the data with which they are provided, they develop their own responses to the questions presented.
Generative AI is just one of many AI implementations, but the focus of this paper is limited to the potential for large networks to create new works as a result of their synthesis of the information they are provided. As explained in this article, generative AI is fundamentally different than other forms of AI because it reflects a new form of automation in which the creative output upon which all societies depend may now be generated by narrow AI systems that do not actually understand what they are producing.[10] The power of a generative AI to compete with humans to tell stories, make music, and create art would trigger significant upheavals across a multitude of industries; to unleash that power to a machine not yet capable of assessing its own creative work makes those implications pale in comparison.
At this stage, most of the AI innovation remains steered toward business.[11] AI systems are excellent at automating routine tasks, while assuring that rote procedures are followed consistently time after time. The systems identify patterns in data that help identify customer interests and behaviors, increasing productivity and profitability. The systems also have the potential to augment the tasks that humans find tedious, increasing workforce satisfaction.[12] Even though these new systems remain error prone, companies are jostling to be first with practical applications. In March 2023, Salesforce “introduced Einstein GPT, an A.I. system that it hopes will help salespeople, marketers, and customer service agents do their jobs more efficiently, as well as detailed plans to integrate the chatbots into the Slack messaging system.”[13]
Salesforce CEO Marc Benioff also identified uses for generative AI in Slack, the business messaging app, describing the impact of generative AI as transformative for commerce. “The world is experiencing one of the most profound technological shifts with the rise of real-time technologies and generative AI….. This comes at a pivotal moment as every company is focused on connecting with their customers in more intelligent, automated, and personalized ways.”[14]
Salesforce is not alone. “Every industry that requires humans to create original work—from social media to gaming, advertising to architecture, coding to graphic design, product design to law, marketing to sales—is up for reinvention.”[15] And yet, the problems have barely been identified and certainly not addressed.
Industry leaders likely hope that the rapid innovations mean that many of the problems will be resolved by the new technology. While the rapid iteration will resolve some of the challenges discussed below, it will only exacerbate others. The anticipated technical improvements may not help with the legal liabilities or the social upheaval. Moreover, in the long run, the potential for generative AI to create a disruptive force to labor markets, cultural institutions, and social norms will go well beyond the sales and marketing concerns being discussed in boardrooms today.
[1] By Jeran Wittenstein, A Factual Error by Bard AI Chatbot Just Cost Google $100 Billion, Time (Feb. 9, 2023), https://time.com/6254226/alphabet-google-bard-100-billion-ai-error/ (“Google parent Alphabet’s shares tumbled 7.7% on Wednesday after concerns surfaced about the competency of Bard, the ChatGPT rival it unveiled on Feb. 6. The selloff continued on Thursday …. The rout has erased about $170 billion in market value.”).
[2] Sundar Pichai, An important next step on our AI journey, Google Message from our CEO (Feb. 6, 2023), https://www.blog.google/technology/ai/bard-google-ai-search-updates/. See Tristan Bove, A robot’s $100 billion error: Alphabet shares tank after its ChatGPT rival makes a mistake in its very first ad, Fortune (Feb. 8, 2023), https://fortune.com/2023/02/08/google-bard-ai-mistake-ad-stock-price-market-cap/.
[3] See Very Large Telescope, European Southern Observatory, https://www.eso.org/public/teles-instr/paranal-observatory/vlt/ (last visited March 9, 2023) (“The VLT has stimulated a new age of discoveries, with several notable scientific firsts, including the first image of an extrasolar planet (eso0428), tracking individual stars moving around the supermassive black hole at the centre of the Milky Way (eso0846), and observing the afterglow of the furthest known Gamma-Ray Burst.”). See Bove, supra note __ (“But the very first image of an exoplanet was captured by the Very Large Telescope, a ground-based array in Chile, in 2004 and confirmed as an exoplanet in 2005, according to NASA—long before James Webb’s 2021 launch. James Webb, however, is being used to identify and catalog exoplanets.”).
[4] See Kif Leswing, Meet the $10,000 Nvidia chip powering the race for A.I., CNBC (Feb. 23 2023), https://www.cnbc.com/2023/02/23/nvidias-a100-is-the-10000-chip-powering-the-race-for-ai-.html (“Nvidia takes 95% of the market for graphics processors that can be used for machine learning, according to New Street Research. … AMD and Intel have competing graphics processors, and big cloud companies like Google and Amazon are developing and deploying their own chips specially designed for AI workloads.”).
[5] See Bern Elliott, Why is ChatGPT Making Waves in the AI Market?, Gartner (Dec. 8, 2022), https://www.gartner.com/en/newsroom/press-releases/2022-12-08-why-is-chatgpt-making-waves-in-the-ai-market (“Artificial intelligence (AI) research and deployment company OpenAI recently announced the official launch of ChatGPT, a new model for conversational AI. According to OpenAI, the dialogue provided by this platform makes it possible for ChatGPT to “answer follow-up questions, admit its mistakes, challenge incorrect premises and reject inappropriate requests.”).
[6] Bove, supra note __.
[7] See Rob Toews The Next Generation Of Large Language Models, Forbes (Feb. 7, 2023), https://www.forbes.com/sites/robtoews/2023/02/07/the-next-generation-of-large-language-models/?sh=a6a71bd18dbc.
[8] Id. See also Lee, supra note __; Wang, supra note __ at 588 (“It is now possible using GANs to generate photorealistic object images such as birds and faces, generate indoor or outdoor scenes, translate images from a source domain to the target domain, generate high-definition images from low-definition images, and so on.”).
[9] AI Generated Books are Flooding Amazon Kindle Store, ODSC (Feb. 23, 2023), https://opendatascience.com/ai-generated-books-are-flooding-amazon-kindle-store/. See also These authors are using ChatGPT to write books and sell them on Amazon, N.Y. Post (Feb. 23, 2023), https://nypost.com/2023/02/21/chatgpt-launches-boom-in-ai-written-e-books-on-amazon/ (“There were over 200 e-books in Amazon’s Kindle store as of mid-February listing ChatGPT as an author or co-author, … [a]nd the number is rising daily. There is even a new sub-genre on Amazon: Books about using ChatGPT, written entirely by ChatGPT.”).
[10] See generally Oladeji M. Tiamiyu, The Impending Battle for the Soul of ODR: Evolving Technologies and Ethical Factors Influencing the Field, 23 Cardozo J. Conflict Resol. 75, 88 (2022) (“The role of AI in art helps to provide a framework for the possibilities of AI in ODR.”); Molly K. Land & Jay D. Aronson, Human Rights and Technology: New Challenges for Justice and Accountability, 16 Ann. Rev. L. & Soc. Sci. 223, 223 (2020); Gia Jung, Do Androids Dream of Copyright?: Examining AI Copyright Ownership, 35 Berkeley Tech. L.J. 1151, 1151 (2020); Harry Surden, Artificial Intelligence and Law: An Overview, 35 Ga. St. U. L. Rev. 1305 (2019); Sonia K. Katyal, Private Accountability in the Age of Artificial Intelligence, 66 UCLA L. Rev. 54 (2019); Frank Pasquale, A Rule of Persons, Not Machines: The Limits of Legal Automation, 87 Geo. Wash. L. Rev. 1 (2019); Yavar Bathaee, The Artificial Intelligence Black Box and the Failure of Intent and Causation, 31 Harv. J. L. & Tech. 889 (2018).
[11] Saarthak Bakshi, Why Artificial Intelligence Is Taking The World By Storm, Business World (June 29, 2019), https://www.businessworld.in/article/Why-Artificial-Intelligence-Is-Taking-The-World-By-Storm/29-06-2019-172451/.
[12] Id. (“Artificial Intelligence helps in directing and locating the entrepreneur to their potential customers, … maintain the several business activities like sales, the flow of cash, inventory and various other records…, assists in better problem-solving, … involves no errors and faults in its functions…., [and] actually reduces the amount of pain that humans have to take….”).
[13] Chris Morris, Salesforce is joining the ChatGPT frenzy by launching an OpenAI-powered tool on Slack, Fortune (March 7, 2023), https://fortune.com/2023/03/07/salesforce-artificial-intelligence-slack-einstein/.
[14] Id.
[15] Sonya Huang, Pat Grady, and GPT-3, Generative AI: A Creative New World, Sequoia (Sep. 19, 2022), https://www.sequoiacap.com/article/generative-ai-a-creative-new-world/.
[16] See Alex Blake, Microsoft will launch ChatGPT 4 with AI videos next week, Digital Trends (March 10, 2023), https://www.digitaltrends.com/computing/chatgpt-4-launching-next-week-ai-videos/.