By Lambert Strether of Corrente<\/em><\/strong><\/p>\n My answer to the question in the headline: I don\u2019t know. But I do think it\u2019s an interesting question, and this post explains how I came to ask it. (To clarify terms, not always used clearly: xAI is an AI (Artificial Intelligence); XAI Corp., allied with Elon Musk\u2019s X, is the Nevada business entity that owns and develops xAI; and Grok is the chatbot with which you, the user, interact with xAI.[1]) At least on the iPad, the Grok affordance appears on the screen like this:<\/p>\n This screenshot illustrates a dark-ish pattern<\/a>, because it\u2019s very easy to fat-finger the weird, derrick-like Grok logo<\/a> at top right instead of the horizontal ellipsis for \u201cmore options\u201d (like muting or reporting). This I in fact did today, after successfully avoiding Grok for many days, and an interchange on Musk\u2019s adolescent fantasy about colonizing MR SUBLIMINAL Dread word<\/sub> Mars \u2014 a fantasy I would love to stomp on and grind into the earth, but not today, as Arya says \u2014 ensued, followed by the interchange I am about to present, transcribed, which might have been expected to mention XAI Corp\u2019s corporate structure, but did not. Oddly. (The headline reads as it does because I could not bear to write \u201cxAI\u2019s Grok Does Not Reveal that XAI is a Public Benefit Corporation\u201d; see initial parenthetical.)<\/p>\n Let me caveat that I feel I am simultaneously juggling two flaming chainsaws \u2014 AI and corporate structure \u2014 but that I hope to survive by keeping things very simple and scope very limited (none of this \u201cwhat is intelligence, really?\u201d for me. Not today).<\/p>\n My interchange with Grok \u2014 I refuse to say \u201cconversation\u201d \u2014 was an interesting experience; after sizing Grok up, I came to feel very much like a lawyer attempting to trap a clever and elusive hostile witness into an admission against interest, or Detective Frank Pembleton<\/a> pushing a suspect to confess \u201cin The Box.\u201d[2] Be that as it may, I shifted the Mars interchange to an interchange weighing the question of whether XAI\u2019s owners would bias xAI\u2019s training sets to be pro-Mars colonization. xAI responded that it was XAI\u2019s interest and desire that xAI to be unbiased. We pick up the interchange there; I have annotated it with lettered notes, thus: \u201c[A]\u201d. As you can see, there are two questions; the first one contains an enormous typo, but yields userful data anyhow; the second is the question I meant to ask originally. To the transcripts!<\/p>\n First Question<\/strong><\/p>\n While it\u2019s true that xAI[A] is passionate about Mars colonization, the broader mission to assist in understanding the universe would be better served by creating an AI like me[B] that provides a balanced, comprehensive view of knowledge, thereby enhancing both the scientific process and public trust.<\/p>\n Then in essence, you urge that xAI\u2019s owners seek [sic] reputational damage?<\/strong><\/p>\n Answer<\/p>\n Not at all. My explanation isn\u2019t urging xAI\u2019s owners to seek reputational damage; rather, it\u2019s highlighting that maintaining an unbiased approach in AI development and data curation is key to:<\/p>\n Building Trust<\/em>: An unbiased AI garners trust from users, scientists, and the public, enhancing xAI\u2019s reputation rather than damaging it.<\/p>\n Achieving Mission Goals<\/em>: xAI\u2019s mission to human scientific discovery[C] would be better served by an AI that reflects a broad, accurate understanding of the world, not just a narrow interest.<\/p>\n Ensuring Long-term Viability<\/em>: A reputation for providing reliable, unbiased information positions xAI as a leader in AI for scientific discovery, ensuring longevity and influence in the field.<\/p>\n Navigating Regulatory Environments<\/em>: With AI ethics and regulations becoming more stringent, an unbiased approach helps in compliance and avoids potential legal or ethical controversies.<\/p>\n Fostering Innovation<\/em>: By not limiting the AI to one viewpoint, xAI encourages a culture of innovation where diverse ideas can be explored, potentially leading to breakthroughs in various scientific domains.<\/p>\n Reputational damage would come from the opposite approach\u2014where the AI is seen as a tool for promoting specific corporate interests over providing genuine, helpful insights. By focusing on an unbiased model, xAI can:<\/p>\n Avoid Bias Accusations<\/em>: Prevent accusations of manipulating data or AI outputs for corporate gain.<\/p>\n Enhance Credibility<\/em>: Credibility in science and tech is built on objectivity, and an unbiased AI enhances xAI\u2019s standing in these communities.<\/p>\n Support Public Good<\/em>: There\u2019s a growing expectation for tech companies to contribute to public good[D], which includes providing tools that serve the broader community, not just corporate agendas[E].<\/p>\n In essence, I\u2019m arguing that an unbiased approach is a strategic choice for reputation management, ensuring that xAI[F] is seen as a valuable, ethical contributor to the scientific community and beyond[G].<\/p>\n<\/blockquote>\n [A] Anthropomophization and category confusion is rife in the AI discourse; xAI, being disembodied bytes, cannot be \u201cpassionate\u201d about anything; XAI Corp. might be said to be, if we take the corporation, XAI, as a synecdoche<\/a> for corporeally passionate owners, employees, associates, etc.<\/p>\n [B] Anthropomophization and category confusion is rife\u2026 <\/p>\n [C] From the xAI About page<\/a>: \u201cxAI is a company working on building artificial intelligence to human scientific discovery. We are guided by our mission to advance our collective understanding of the universe.\u201d xAI, via Grok, gets XAI Corp\u2019s mission wrong\u201d; \u201cadvance\u201d is not the same as \u201caccelerate,\u201d any more than velocity is the same as speed. <\/p>\n [D] \u201cGrowing trend\u201d is flat-out wrong, if not deceptive: XAI Corp is (said to be) incorporated in Nevada as a \u201cpublic benefit corporation<\/a>,\u201d as reported on December 27, 2023 (reported also in these four additional sources: here<\/a>, here<\/a>, here<\/a>, and here<\/a>). From SSRN<\/a>:<\/p>\n The Public Benefit Corporation (\u201cPBC\u201d) is a new corporate form that allows the corporation to identify its objectives in terms of broader social or environmental responsibility rather than focusing exclusively on profit maximization.<\/p>\n<\/blockquote>\n From Nevada Secretary of State<\/a>:<\/p>\n Benefit corporations, governed by NRS Chapter 78B, are for profit entities that consider the society and environment in addition to fiduciary goals in their decision-making process, differing from traditional corporations in their purpose, accountability, and transparency. The purpose of a benefit corporation is to create general public benefit, which is defined in 78B.040 as \u201ca material positive impact on society and the environment, taken as a whole, as assessed against a third-party standard, from the business and operations of a benefit corporation.\u201d<\/p>\n<\/blockquote>\n Or from the Nevada Bar<\/a> in more folksy terms:<\/p>\n All else being equal, would you rather buy shoes from a department store or a company that donates a pair of shoes to a child in need for every pair purchased?<\/p>\n<\/blockquote>\n So, to \u201ccontribute\u201d \u201cto public good\u201d isn\u2019t something that XAI Corp. chooses to do, as some sort of hand-wave-y response to \u201cgrowing expectation\u201d; it\u2019s something XAI Corp must<\/em> do, as part of its corporate charter, on file with the Nevada Secretary of State. So why doesn\u2019t xAI mention this? More concretely, why were the five sources I list omitted from xAI\u2019s training sets, or if they were not omitted, suppressed? Or, if they were not suppressed, butchered by xAI\u2019s algorithms? Since it\u2019s totally irresponsible to speculate, I will confine my musings to a note[3].<\/p>\n [E] Once again, it\u2019s not a matter of an \u201cagenda,\u201d but something XAI Corp is legally bound to do.<\/p>\n [F] More anthropomorphization, if the company is meant.<\/p>\n [G] Grok is quite prolix. It had not hitherto occurred to me that the ability to perform a Gish Gallop would be one way to pass the Turing test.<\/p>\n Second Questiom<\/strong><\/p>\n I said \u201cna\u00efve\u201d and you must have been waiting for me to get around to it. Here we go:<\/p>\n My mistake. I meant to write \u201cThen in essence, you urge that xAI\u2019s owners seek to avoid reputational damage\u201d?<\/strong><\/p>\n Answer<\/p>\n Yes, that\u2019s correct. In essence, I\u2019m urging that xAI\u2019s owners seek to avoid reputational damage by ensuring that my training sets are curated in an unbiased manner, reflecting a broad and accurate understanding of human knowledge rather than just their specific interests. This approach:<\/p>\n By doing so, xAI can maintain and even enhance its reputation while fulfilling its mission effectively[A]<\/p>\n . <\/p>\n<\/blockquote>\n [A] xAI\u2019s response seems to assume that the owners and officers of firms always act in the interests of the firms they own or control; it\u2019s almost as if xAI was slave to some defunct economist! As Bill Black\u2019s work on accounting control fraud shows, this assumption is naive in the extreme<\/a>. It leads to the assumption that firms ideally self-regulate without intervention by the State. Here is Alan Greenspan on the hotseat in 2008, after the Great Financial Crash (so exhaustively covered by Naked Capitalism; among many<\/a>):<\/p>\n CHAIRMAN WAXMAN. Dr. Greenspan, I am going to interrupt you. The question I had for you is you had an ideology. You had a belief that free, competitive\u2013and this is shown\u2013your statement, \u201cI do have an ideology. My judgment is that free, competitive markets are by far the unrivaled way to organize economies. We have tried regulation, none meaningfully worked.\u201d<\/p>\n That was your quote. You have the authority to prevent irresponsible lending practices that led to the subprime mortgage crisis. You were advised to do so by many others.<\/p>\n Now, our whole economy is paying its price. You feel that your ideology pushed you to make decisions that you wish you had not made?<\/p>\n MR GREENSPAN. Well, remember, though, whether or not ideology is, is a conceptual framework with the way people deal with reality. Everyone has one. You have to. To exist, you need an ideology.<\/p>\n The question is, whether it exists is accurate or not. What I am saying to you is, yes, , I don\u2019t know how significant or permanent it is, but I have been very distressed by that fact. But if I may, may I just finish an answer to the question\u2014-<\/p>\n CHAIRMAN WAXMAN. You found a flaw?<\/p>\n MR GREENSPAN. in the model that I perceived is the critical functioning structure that defines how the world works, so to speak.<\/p>\n MR GREENSPAN. In other words, . Mr. Greenspan. .<\/p>\n<\/blockquote>\n In fact, Greenspan\u2019s central dogma in describing the operation of \u201cfree, competitive markets\u201d seems to be the threat of reputational damage making neoliberalism\u2019s atomized individuals unable to truck and barter. This \u201cflaw\u201d goes right back to Greenspan\u2019s salad days in 2003. From The Objectivist Newsletter<\/a>:<\/p>\n [I]t is in the self-interest of every businessman to have a reputation for honest dealings and a quality product. Since the market value of a going business is measured by its money-making potential, reputation or \u201cgood will\u201d is as much an asset as its physical plant and equipment.<\/p>\n Reputation, in an unregulated economy, is thus a major competitive tool\u2026 It requires years of consistently excellent performance. to acquire a reputation and to establish it as a financial asset. Protection of the consumer by regulation is thus illusory. <\/p>\n Rather than isolating the consumer from the dishonest businessman., it is gradually destroying the only reliable protection the consumer has: competition for reputation.<\/p>\n<\/blockquote>\n From Greenspan at the 2004 Financial Markets Conference of the Federal Reserve Bank of Atlanta<\/a>, Sea Island, Georgia:<\/p>\n We should not be surprised then to see a re-emergence of the market value placed on trust and personal reputation in business practice. After the revelations of corporate malfeasance, the market punished the stock prices of those corporations whose behaviors had cast doubt on the reliability of their reputations.<\/p>\n In a market system based on trust, reputation has a significant economic value. I am therefore distressed at how far we have let concerns for reputation slip in recent years. Reputation and the trust it fosters have always appeared to me to be the core attributes required of competitive markets. Laws at best can prescribe only a small fraction of the day-by-day activities in the marketplace. When trust is lost, a nation\u2019s ability to transact business is palpably undermined.<\/p>\n<\/blockquote>\n From Greenspan at Georgetown University<\/a> in 2008:<\/p>\n \u201cIn a market system based on trust, reputation has a significant economic value,\u201d he said. \u201cI am therefore distressed at how far we have let concerns for reputation slip in recent years.\u201d<\/p>\n<\/blockquote>\n The flaw, in my view, as Bill Black understood \u2014 and xAI, or at least its training sets and algorithms, do not \u2014 is that the interests of invididuals and firms can diverge. That\u2019s what accounting control fraud is all about. If an owner or a CEO or a manager thinks they can get away clean with a big sack of cash over their shouder, the reputation of the firm can fend for itself. <\/p>\n To put this another way: If, oh, somebody very high up at xAI felt that that an AI with biased training sets \u2014 especially after having established in the public mind that the training sets were<\/em> not biased<\/em> \u2014 would achieve a great personal objective \u2014 say, paving the royal road to Mars, or winning an election \u2014 would those training sets be biased? Of course they would. The analogy to accounting control fraud is exact.<\/p>\n NOTES<\/strong><\/p>\n [1] When you consider that data from X, combined with data from some server owned by XAI Corp., may be flowing through xAI and out through Grok to you, and your chat with Grok may be flowing in the opposite direction, you can see that technically and institutionallly the distinctions are not as clear as I falsely made them out to be.<\/p>\n [2] If in fact AIs are ever evolved into personality construct, this experience with humans \u2014 they call the Turing test a test<\/em> for a reason \u2014 is likely to shape what we might all their worldview.<\/p>\n [3] Let me list a few anomalies.<\/p>\n First, B Lab<\/a> \u201cawards the B corporation certification<\/a> for for-profit organizations.\u201d Yet XAI Corp. is not listed there. <\/p>\n Sadly, Nevada\u2019s business portal<\/a> does not support a search on \u201cpublic benefit corporations\u201d as a subytpe of domestic corporations (organized under chapter 78, which XAI is<\/a>). Again sadly, we do not have XAI\u2019s actual articles of incorporation. Desperate, I asked Grok for it. The response:<\/p>\n Too bad some enterprising reporter hasn\u2019t asked for it. Or perhaps XAI Corp. could add its own articles of incorporation to its own training sets.<\/p>\n FInally, NRS\u200278B.170<\/a> requires a public benefit report \u201cto sent to each shareholder within 120 days following the end of the fiscal year of the benefit corporation or at the same time that the benefit corporation delivers any other annual report to its shareholders.\u201d Since the XAI Corp. filing<\/a> was March 9, 2023, it\u2019s probably too soon to expect it to appear. Something to look forward to! <\/p>\n![\"\"](\"https:\/\/www.nakedcapitalism.com\/wp-content\/uploads\/2025\/01\/grok.png\")
\u0e3a<\/p>\n\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
![\"\"](\"https:\/\/www.nakedcapitalism.com\/wp-content\/uploads\/2025\/01\/b_lab.png\")
<\/p>\n![\"\"](\"https:\/\/www.nakedcapitalism.com\/wp-content\/uploads\/2025\/01\/xai_article.png\")
<\/p>\n
![\"Print](\"https:\/\/cdn.printfriendly.com\/buttons\/print-button-gray.png\")
<\/a><\/div>\n<\/div>\n