Saturday, April 1, 2023

Disruption versus Destruction.

Today's edition of the New York Times caught my eye with an article by Cade Metz regarding Sam Altman, the CEO of OpenAI, and their chatbot, ChatGPT. The article weaves between Altman's biography, the product, and the nature of the firm. The article references Altman's belief in "effective altruism," a rational approach to maximizing global impact through evidence-based decision-making. The article's almost treats the life story of a CEO as a political biography, as a source to establish the legitimacy of the company and the technology.

A tech firm CEO, much like a head of state, serves as a unifying force that seeks to transcend divisions and foster a sense of identity and shared values around the company. Sam Altman's "effective altruism" is here like the "dignified" aspect of a monarchy, providing reassurance to the public and calming concerns about the impact of AI. Meanwhile, the "efficient" aspect of a tech firm, responsible for developing and implementing technology, operates on business principles necessarily different from the legitimizing elements of the firm's leadership.

While the New York Times article illustrates the role of a technology CEO in shaping public perception, it is crucial to remember that the efficient aspect of OpenAI is far different than the personality of Sam Altman. Effective altruism has little to do with the actual impact of the technology. It is vital to focus on the potential risks of disruptive new technologies, as their misuse in authoritarian hands has been destructive. "Effective altruism" could very well be serving as a distraction here, rather than a guarantee of responsible AI development.

Friday, March 31, 2023

@TheEconomist and @duncanrobinson on #RoaldDahl.

Puffin, the publisher of Roald Dahl (Charlie and the Chocolate Factory, James and the Giant Peach, etc.), recently edited some of Dahl’s works for sensitivity, removing words such as "fat," "flabby," "ugly," and "Kipling." This act, which sparked a backlash, it is argued by Bagehot, the Economist’s British politics columnist (Economist articles are traditionally unsigned, but this one is by Duncan Robinson) is part of a broader trend in British publishing, where books are being censored or dropped, and sensitivity readers are employed to ensure adherence to modern morals.

Roald Dahl, 1954. 

While the right to prepare derivative works is at the core of copyright, the editing of Dahl's work by Puffin, a Penguin imprint, is argued to be just one symptom of a deeper issue in the publishing world. Making an impressive leap, Bagehot contends the argument that suppression of speech is only a problem in totalitarian states fails to recognize the "veiled censorship" in British publishing. There is an orthodoxy that right-thinking people are expected to accept without question, and resistance to the same leads (at least in the mind of the columnist) to being silenced with surprising effectiveness.

Publishers, in an attempt to look likable, often panic and preempt offense, leading to the removal or editing of content. However, this nervousness and desire to look nice can have nasty results, as it stifles creativity and prevents important discussions from taking place. The observation that the publishing industry is susceptible to peer pressure sounds in truth, as any observer of media generally is keenly aware of herd effects and the power of groupthink in the industry.

Where the columnist goes awry (and this is perhaps to be expected for a print journalist) is that there are a variety of means where unorthodox ideas can reach a broader audience. If anything, the rise of misinformation through alternative channels presents far broader problems for democracy than were ever perceived twenty to thirty years ago. Editors and publishers are not all wrong, and sometimes, they are even right as a group. I feel I can understand both the germ of the argument and the (veiled?) frustration of the writer given the unique power of publishing in certain professional and cultural circles. Sending ideas out into the world in book form is a form of professional and social recognition oftentimes far exceeding the economic import of such an activity.

Given the foregoing, I think the argument advanced, and particularly the vignette of how the suppressive mechanism works, is powerful. “What is striking is how apparently mild the sanctions are for speaking out … what really terrifies you is that your colleagues will think a little less of you. Most people do not require the threat of being burned at the stake to shut them up; being flamed by their peers … is more than enough.” Nudges in favor of conformity are often powerful precisely due to their superficial innocuousness – a timeless observation if there ever was one.