ethics

Bilingual

AI Everywhere, Regulation Nowhere? Two Years of AI Policy in the UK
2 months ago
- UK government prioritized AI adoption over regulation, resulting in a lack of dedicated AI regulatory framework.
- The UK refused to sign the Paris AI Safety Summit declaration, aligning with the US deregulation approach.
- Government initiatives like the AI Opportunities Plan and Parlex chatbot were criticized as hyperbole and ineffective.
- Massive government contracts awarded to US AI firms (Palantir, OpenAI) faced scrutiny over overblown benefits and phantom investments.
- Inappropriate uses of AI surged, including facial recognition bias, police misconduct, and flawed age checks for child migrants.
- AI regulation efforts stalled; a private member's bill in the Lords lacked government backing and went nowhere.
- The UK is now years behind the EU in AI regulation, lacking powerful tools like the Digital Services Act or a centralized regulator.
Questions and answers on AI and verification: follow-up to my May ACM Tech Talk
2 months ago
- The author gave an ACM Tech talk on 'Software Verification in the Age of Artificial Intelligence' that was well-attended.
- Due to time constraints, many questions were left unanswered; the author promised to address them.
- After finishing his book 'AI for Smarties: Understanding AI', the author is now returning to answer the pending questions.
- The questions cover a wide range of topics including ethics, safety, formal methods, AI agents, copyright, and the impact on software engineering.
- The author invites further questions and comments via the blog's comment section and looks forward to continuing the discussion.
For a Business, Should Making (Extra) Money Be Incidental?
2 months ago
- Unchecked growth in business is celebrated, but in medicine it would require a doctor; a business should primarily serve the public, with profit as incidental.
- Vicky, a Greek restaurant owner in Chicago, prioritizes quality and affordability over profit, using ripe tomatoes, extra virgin olive oil, and fresh organic meat.
- She avoids hormone-pumped chicken and advises customers to order less to avoid waste, showing service over sales maximization.
- In Japan, customers are treated as 'okyakusama' (like gods), while in America, profit is god and customers are reduced to consumers exploited by psychological and legal tricks.
- Businesses should help satisfy desires without exploiting people; supporting 'Vickies'—service-oriented businesses—is a meaningful action for individuals and the economy.
February 2021 Gwern.net Newsletter
2 months ago
- Gwern.net newsletter for February 2021 covers AI scaling, genetics, statistics, politics, psychology, technology, economics, and philosophy.
- AI section highlights advancements in neural text generation, language model fine-tuning, prompt programming, and brain-computer interfaces.
- Genetics topics include heritability of traits, recent evolution, and polygenic burden impacts on health and cognition.
- Statistics/meta-science discusses lessons from history, microCOVID risk calculator, and critiques of modern mathematics education.
- Politics/religion covers fashion cycles, lunar pandemic quarantine critique, and Lysenkoism in Russia.
- Psychology/biology focuses on semaglutide for weight loss, bat immune defenses, and placebo effects in psychedelic microdosing.
- Technology section mentions wringing gauge blocks as a mysterious physical phenomenon.
- Economics analyzes renewable energy cost reduction, IQ and trading performance, and genetic endowments in wealth inequality.
- Philosophy questions ethics experts' behavior, with studies showing they may not be more ethical than others.
- Key themes include scaling laws, biological plausibility of backprop, and the role of placebo in psychology.
Pluralistic: "Rights for robots" and the AI slavery fantasy (10 Jul 2026)
2 months ago
- The AI bubble is driven by bosses' desire to replace workers with software, fueled by a fantasy of a world without people except for slaves.
- AI often hides human labor, such as low-wage overseas workers acting as 'humans in the loop,' treated without moral consideration.
- The 'rights for robots' thought experiment inadvertently validates the idea that AI are slaves, reinforcing the AI sales pitch.
- Extending personhood to artificial constructs like corporations has been harmful, unlike extending rights to nature, which benefits humans.
Gatekeepers, Luddites, Haters and AI-Generated Maps
2 months ago
- The author opposes generative AI in cartography, rejecting accusations of being a gatekeeper, Luddite, or AI hater.
- AI-generated maps are promoted by professionals, but the author, a non-mapmaker, refuses to use them because the AI does the work, not the user.
- Key issues: AI-generated maps lack copyright protection, contain serious inaccuracies (e.g., maps with two Ohios), and are shared without disclaimers, leading to ethical concerns.
- AI adoption in other fields has led to mass layoffs and precarious work, not leisure; professionals should be Luddites opposing poor conditions.
- Comparing generative AI to typewriters is flawed; the author likens it to cocaine—addictive and impairing creative thinking.
- Broader costs include environmental damage, plagiarism lawsuits, bubble economics, and public fury over AI replacing human interaction.
AI in Linux
2 months ago
- Linus Torvalds supports using AI tools like LLMs for code authoring and review in Linux kernel development, with over 1,200 commits having an 'Assisted-by' tag.
- The new code review tool Sashiko generates AI-driven reviews, forcing contributors to interact with AI and sparking debate about mandatory AI usage.
- Linus dismisses ethical and political arguments, insisting Linux focuses solely on technology, but the author argues Linux is inherently political due to licensing choices and the practical impossibility of forking.
- AI externalities include rising hardware prices, carbon emissions, and energy consumption, with negative impacts on people including kernel contributors.
- The author criticizes Linus for legitimizing AI companies and calls for honest discussion of moral and ethical dimensions rather than narrow technical focus.
Ethics Is the Skill That Matters Most Now
2 months ago
- Skills like writing, coding, analysis, and design are being commoditized by AI, shifting value from capability to judgment about what should be done.
- AI amplifies whoever holds it; the bottleneck is intent, not capability, and bad intent was historically throttled by incompetence.
- Ethics is a trainable skill involving noticing moral dimensions, reasoning under conflict, owning second-order effects, and acting against pressure.
- AI raises ethical stakes through scale (mass replication of errors), speed (faster than review), and distance (layers diffusing responsibility).
- Delegating ethics to AI fails because human judgment must specify values; moral skills atrophy when not exercised.
- Political systems filter for traits like lying and manipulation, rewarding unethical leaders at a time when ethics is critical.
- Practical habits include considering affected parties, designing for misuse, speaking up early, and treating justifications as warnings.
- Intelligence is automated, but judgment is not; ethics is the skill that makes capability worthwhile and cannot be outsourced.
FCC Officials Took $$$ Gifts from Paramount as Company Needed Approval for Deals
2 months ago
- FCC commissioners have accepted expensive Kennedy Center gala tickets from CBS/Paramount, raising ethical concerns.
- Commissioners Olivia Trusty and Brendan Carr accepted tickets worth thousands of dollars after voting on or overseeing Paramount-related mergers.
- Ethics experts argue this creates a clear conflict of interest, compromising FCC impartiality in pending merger decisions.
- Federal rules generally prohibit gifts from regulated entities, but exemptions for 'widely attended gatherings' are controversially applied.
- The FCC's ongoing review of the Paramount-Warner Bros. Discovery merger could be legally challenged due to these ethics issues.
- Some commissioners, including Carr, have a history of accepting such gifts, totaling over $260,000 from CBS/Paramount since 2016.
- Potential recusals could deadlock the FCC, as it currently has only three commissioners, affecting merger approval processes.
- Legal experts warn that ethics violations might undermine the legitimacy of FCC decisions in court.
2014: DeepMind reportedly sold to Google on conditions: no military use
2 months ago
- The article discusses DeepMind's transition from a company with safeguards against military use to one accepting Pentagon contracts.
- Andreas Kirsch, an employee at Google DeepMind, highlights public reports from 2014 and 2026.
- In 2014, DeepMind was sold to Google with conditions including no military use and independent oversight.
- By 2026, it had a Pentagon contract for "any lawful government purpose", indicating the safeguards did not survive intact.
International chess federation sanctions Kramnik
3 months ago
- The FIDE Ethics & Disciplinary Commission (EDC) found former World Champion Vladimir Kramnik responsible for multiple violations of the FIDE Ethics and Disciplinary Codes.
- The violations included breaches related to dignity, respectful treatment, bullying, cyberbullying, psychological abuse, and failure to cooperate with a Fair Play Commission investigation.
- Several other charges, such as those concerning integrity, honesty, and reputational harm to FIDE, were dismissed due to insufficient evidence.
- The EDC emphasized that combating cheating is a top priority but must be handled through confidential procedures with proper evidence, not public accusations.
- Kramnik received a two-year ban from FIDE competitions, with the final 12 months suspended for a three-year probation, resulting in an active one-year suspension if no further breaches occur, plus 12 months of unpaid community service.
AI children's books, body horror edition
3 months ago
- AI-generated children's books lack uniqueness and exhibit sameness despite improvements in models.
- The author purchased an Amazon bestseller AI children's encyclopedia, noting targeting reasons: high sales, gift purchases by non-readers, and no intellectual property risks.
- The book contained disturbing body horror imagery, including mutated creatures and unsettling scenes, raising concerns about its impact on children.
- AI-generated content is widespread on platforms like Amazon, with potential real sales despite fake reviews, highlighting regulatory needs.

ethics

AI Everywhere, Regulation Nowhere? Two Years of AI Policy in the UK

Questions and answers on AI and verification: follow-up to my May ACM Tech Talk

For a Business, Should Making (Extra) Money Be Incidental?

February 2021 Gwern.net Newsletter

Pluralistic: "Rights for robots" and the AI slavery fantasy (10 Jul 2026)

Gatekeepers, Luddites, Haters and AI-Generated Maps

AI in Linux

Ethics Is the Skill That Matters Most Now

FCC Officials Took $$$ Gifts from Paramount as Company Needed Approval for Deals

2014: DeepMind reportedly sold to Google on conditions: no military use

International chess federation sanctions Kramnik

AI children's books, body horror edition