What if, for example, commercial use of agentic AI was governed by a contract dynamically generated and negotiated by AI systems? The contract could specify different risk sharing terms based on the inherent risks of the activity under consideration, themselves modeled quantitatively by teams of AI actuaries and economists. It could consider the likelihood of third-party harms (harms caused by an AI system to a bystander uninvolved in the contract) and include provisions for compensating third parties.

we will not successfully design an AI governance regime from the top down; we will discover it, and one essential tool for doing so will be the resolution of concrete, realized harms.

It is a system that shapes what you do and do not eat, what kind of medicines you can and cannot take, and even the educational opportunities available to children. It is a system that profoundly affects nearly every product, public space, and even many services you encounter—and prevents many products, public spaces, and services from existing altogether.

If the Court finds in favor of the plaintiffs, it will establish a beachhead for liability when a platform uses an algorithm to surface content to users.

In some cases, liability may seem desirable. An LLM could make it easier to break the law, such as if it provides instructions on how to build a bomb, or manufacture and distribute illegal drugs. It could make it easier for a student to cheat. It could use racist or misogynistic language, or help fuel disinformation campaigns. Making LLMs liable for problematic content they generate in response to a user prompt will push engineers to design the products to minimize the likelihood they produce harmful content.

The relevant question will be whether LLMs “develop” content, at least “in part.” It is difficult to imagine that a court would find otherwise if an LLM drafts text on a topic in response to a user request or develops text to summarize the results of a search inquiry (as ChatGPT can do). In contrast, Twitter does not draft tweets for its users, and most Google Search results simply identify existing websites in response to user queries.

New crack – Third Circuit’s TikTok decision. A recent Third Circuit ruling broke from that pattern, suggesting a platform’s algorithmic recommendations can give rise to liability. In Anderson v. TikTok, Inc. (2023), the estate of a 10-year-old girl sued TikTok after she died attempting the viral “Blackout Challenge” allegedly promoted to her by TikTok’s “For You” algorithm. The district court had dismissed the case under Section 230, but the Third Circuit reversed in a landmark decision, holding that TikTok could potentially be held liable for its role in actively recommending dangerous content. The court reasoned that TikTok “is not merely hosting third-party content but actively recommending specific content to users,” effectively engaging in its own form of expression or conduct. By curating and targeting the challenge videos to a vulnerable child, TikTok’s algorithm crossed the line from passive intermediary to active participant, in the Third Circuit’s view. While Section 230 protects publishing third-party content, it “may not extend to the algorithms [platforms] use to curate and promote that content”. One concurring judge argued forcefully that platforms should not be immune when their proprietary algorithms push deadly content to kids, criticizing the “causal indifference” and lack of accountability under the old 230 paradigm.

In contrast, a Section 230defense sometimes is so obvious, backed by decades of precedential defense wins,that defendants will treat victory as one hundred percent assured. They are lesslikely to settle those cases, even for a fraction of defense costs

Also, complex litigation can divert substantial managerial and organizationalattention and mindshare from maintaining or enhancing the service. Thus, the abilityof a defendant to resolve a case on a motion to dismiss (and avoiding expensivediscovery) protects small and low-revenue Internet services, which in turn enhancesthe richness and diversity of the Internet ecosystem.5

Where the incentives for self-restraint are mitigated, there will be more speech andconduct, and relatively more of it will be harmful or illegal

They are, in other words, free-speech absolutists. Thisis not our position, though we are staunchly defensive of free-speech rights and count the prospectof lost opportunities for user-generated speech as a significant potential cost of any limitation onSection 230 immunity.13 Depending how speech is weighted in the calculus, some may concludethat the benefits of our proposed approach are not worth the costs. That is a tenable position

n another recent case, Force, the Second Circuit considered a claim thatFacebook unlawfully assisted Hamas in terrorist attacks coordinated onFacebook by promoting “personalized” content.83 The court held thatFacebook’s use of algorithms to “match” content to users “based on objectivefactors applicable to any content” did not constitute development of thecontent where the content was “materially unaltered.

However, it should be noted that in cases where Generative AI outputs are so close to the source material that the Generative AI model does not materially contribute to the creation of the illegal content and is acting solely in the capacity of an interactive computer service, currently applicable case law indicates those outputs could still receive Section 230 immunity. To this end, there is existing precedent in which courts have held that algorithmic recommendations or use of algorithms even in the proliferation of potentially illegal content was protected under Section 230 immunity.

On the other hand, if the model were to return a wholly new original output that contained potentially defamatory or illegally discriminatory content in response to an otherwise legal prompt, the output would likely be a material contribution to the allegedly illegal nature of the content.

Given that Generative AI systems engage in a wide breadth of functions, some of which may not involve the creation of original content or might be significantly shaped by user prompts, determining whether the system is an “information content provider” with respect to particular content, and thus outside the scope of Section 230 immunity, would likely vary on a case by case basis.

This is just one of the 70 association studies Haidt cited, but almost all of them suffer from the issues tabulated above. Not all of these problems were in all of the studies, but none of the 68 had a clear, strong result demonstrating above-normal depression levels of heavy social media users based on reliable data and robust statistical methods.

"More and more kids are getting connected with treatment. And those treatments are specific for suicide and suicide prevention."

One way to characterize the Keynesian approach (see below) is that it gives almost exclusiveimportance to the first-round effect by putting primary emphasis on flows of spending rather thanon stocks of assets

"The issues of a Government paper, evenwhen not permanent, will raise prices; because Governments usually issue their paper inpurchases for consumption. If issued to pay off a portion of the national debt, we believe theywould have no effect"

In this respect the correct analogy is with the demand for, say,land, which, like money, derives its value from the flow of services it renders but has a purchaseprice and not merely a rental value.

From this point of view, it is important to distinguish between ultimate wealth holders, to whommoney is one form in which they choose to hold their wealth, and enterprises, to whom money isa producer’s good like machinery or inventories

The quantity theory of money takes for granted, first, that the real quantity rather than thenominal quantity of money is what ultimately matters to holders of money and, second, that inany given circumstances people wish to hold a fairly definite real quantity of money

A different way to express the real quantity of money is in terms of the time duration of the flowof goods and services the money could purchase

The United Kingdom’s Online Safety Act (OSA) went into effect July 25,

By including “misleading” speech, the misinformation definition covers an almost immeasurable amount of speech, much of which is just speech that a given person disagrees with.

Under this definition, countless examples of journalism, academic research, and other forms of communication might fairly be labeled as misinformation because of how a story is framed, what evidence is included or excluded, how someone expresses their opinions or policy positions, etc., or any other array of imperfections that exist in natural human communication’s inevitable subjectivity and necessary choices

Although the core components of Plaintiffs’ final judgment remain, a few significantitems have changed. As detailed further below, Plaintiffs no longer seek the mandatorydivestiture of Google’s AI investments in favor of a prior notification for future investments andhave modified the ads syndication remedy to focus on parity, transparency, and control, whileremoving the query volume limitation and implementing marginal cost pricing only ascontingent relief if Plaintiffs’ other remedies are not effective at restoring competition.

To remove barriers to entry and erode Google’s unlawfully gained scale advantages,Section VII requires Google to syndicate (subject to certain restrictions) its search results,ranking signals, and query understanding information for 10 years. See Mass., 373 F.3d at 1218(disclosure of APIs “represent[ed] a reasonable method of facilitating the entry of competitorsinto a market from which Microsoft’s unlawful conduct previously excluded them” (internalquotation omitted)). The RPFJ only requires Google to syndicate queries that originate in theUnited States. See RPFJ ¶ VII(B)

Even if it were technically feasible, a structural breakup of a natural monopoly search engine would be a disaster. It would impose two or more suboptimal products on consumers

Finally, even if we assume them to be methodologically flawless, the studies only found small effect sizes. Meaning, that even if violent video games did increase aggression, the effect would be quite small.

This investigation therefore highlights two intrinsic problemsconfronting behavioural scientists using large-scale social data.First, large numbers of ill-defined variables necessitate researcherflexibility, potentially exacerbating the garden of forking pathsproblem: for some datasets analysed there were more than a trilliondifferent ways to operationalize a simple regression

heterogeneous results, banning mobile phones could be a low-cost way for schools to reduceeducational inequality

Our datasuggest that interventions to reduce phone/social mediatime to positively influence adolescent mental wellbeingare plausible, but that both in-school and outside ofschool use should be considered in tandem.

It is the government’s burden to prove that thelaw serves a compelling government interest and usesthe least restrictive means to achieve that interest.

he findings have both practical and theoretical implications. Prac-tically, the social media fact-checks alone may not be enough of adeterrent to stop the spread of misinformation unless they are accom-panied by further sanctions.

Research on the effects of fact-checking labels provides mixed results: Some found such labels effectively reducing perceived accuracy of false information (Pennycook et al., 2020) and willingness to share such content (Nekmat, 2020; Yaqub et al., 2020), but others found little effects of labels on perceived credibility, sharing intention, or engagement (Bradshaw et al., 2021; Oeldorf-Hirsch et al., 2020; Papakyriakopoulos & Goodman, 2022).

Our research shows how the ‘community notes’ model –which combines automation with the ‘wisdom’ of user crowds– is ill-equipped to address hate-fuelled, harmful narratives often deeply entwined with disinformation. Rather than being a more empowering and comprehensive system as Meta claims, its Community Notes will likely mirror and extend the divisive, ideological leaning of its CEO, erecting another wrestling ring for the battle over ‘facts’ that favors the political right in an online environment where marginalized communities have fewer protections than ever before.

Improving youth experiences online requires a holistic approach. Using a mix of alternative methods to improve youth—and general user—safety online may more effectively and directly address concerns about access to age-inappropriate materials and the negative impact of online spaces. Ultimately, age verification is no substitute for privacy protections and increased user transparency and control

Chinese labs are more transparent than their government in part because they want to create an ecosystem of firms centred on their AI. This has some commercial value, in that the companies building on the open-source models might eventually be persuaded to buy products or services from their creators. It also brings a strategic benefit to China, in that it creates allies in its conflict with America over AI.Chinese firms would naturally prefer to build on Chinese models, since they do not then need to worry that new bans or restrictions might cut them off from the underlying platform. They also know they are unlikely to fall foul of censorship requirements in China that Western models would not take into account. For firms like Apple and Samsung, eager to build AI tools into the devices they sell in China, local partners are a must, notes Francis Young, a tech investor based in Shanghai. And even some firms abroad have specific reasons for using Chinese models: Qwen was deliberately imbued with fluency in “low-resource” languages such as Urdu and Bengali, whereas American models are trained using predominantly English data. And then there is the enormous draw of the Chinese models’ lower running costs.

Thus, while it may be the case that only one firm may be producing in a particular market, this does not necessarily imply, as Demsetz and Kirzner highlight, the absence of rivalrous competition among entrepreneurs for ownership of such a firm. 

Para Zambrano "el gobierno no tenía otra opción" si quería mantener sus calificaciones crediticias.

de los $us 2.300 millones. Es una meta reducida y con escaso o nulo impacto sobre la lucha contra este flagelo que tiene tendencia creciente en el contexto del Covid-19”, insinuó Blazicevic.

Llamó al Gobierno a proteger la industria nacional aumentando los controles

More than 100 years ago, the philosopher John Stuart Mill provided two of the best reasons for why Peterson should be supported in his struggle. The first is that Peterson’s arguments could be true. This is a distinct possibility because of the political climate in which trans issues are currently discussed. There is a reluctance to question the arguments of the “positive space” lobby, and such groupthink often leads to poorly thought out arguments

a study from Rutgers University found that “white officers are no more likely to use lethal force against minorities than nonwhite officers,” in the words of lead researcher Charles Menifield.

the researchers postulate that this may be due to “simple overlap between officer and county demographics.” Police departments in areas with greater numbers of ethnic minorities tend to have a more diverse police force.

Indeed, despite decades of aggressive policing and the tens of billions of dollars allocated to law enforcement every year, proponents of defunding the police assert, the United States still has more murders, rapes, and robberies proportionately than countries that also enjoy much lower levels of police violence. "Does this spending make the country safer than its peers?" Lowrey asked. 

The police fatally shot nine unarmed blacks and 19 unarmed whites in 2019," Mac Donald wrote. "Assuming a comparable number of victims last year, those nine unarmed black victims of police shootings represent 0.1% of all African-Americans killed in 2019. By contrast, a police officer is 18½ times more likely to be killed by a black male than an unarmed black male is to be killed by a police officer."

police patrols, arrests, fines, and incarceration as methods of combatting crime, proponents of defunding the police contend, it would be better to shift government resources to social and mental health services, education, housing assistance, welfare, and other programs that, they assert, can prevent crime and serve the public much more effectively and safely than police can.

Imagine a world without Flo-Jo, without Venus or Serena Williams, without modern-day Olympian runner Allyson Felix. That is the world brought about by advocates of transgender-identified biological males competing with women in sports: contests so lopsided that our greatest female athletes never qualify for the Olympics, never make it to Wimbledon—let alone astound us there, with feats of the gods.

The other side of the argument is no argument at all. It is fear. Woke mobs have managed to scare good people into silence. Ordinary Americans are afraid to point out even manifest injustice when they see it. Too many are stricken by the threat of being falsely branded some flavor of "phobe.

The only other option comes with too steep a price. To force young women to compete with male-bodied athletes will bring about the collapse of women's sports

These "organizational" effects of male puberty are profound, and they are permanent: larger hearts and lungs, more oxygenated blood, more fast-twitch muscle fiber, greater upper-body muscle mass, greater lower-body muscle mass, greater bone density

If this were not 2020—if Americans were not terrified to acknowledge the plain truths in front of them—this discussion would be confined to its more natural domain: high school auditoriums, where boys and girls in glasses and blazers can earn high marks for casuistry

Research Questions and Hypothesis

Thompson et al. (2011) gave a first hint as to why previous results were so mixed. They revealed that music characteristics like tempo and intensity have an influence on learning outcomes: only soft fast music had a positive influence, whilst loud fast as well as soft slow or loud slow music hindered learning.

Results of studies investigating the relationship between background music and learning outcomes are varied. While some studies found no effect of background music (e.g., Moreno and Mayer, 2000; Jäncke and Sandmann, 2010) others found that it negatively impacted learning outcomes [e.g., Furnham and Bradley, 1997; Randsell and Gilroy, 2001; Hallam et al., 2002 (study 2)]. Further studies report that it has a positive impact [e.g., Hallam et al., 2002 (study 1); de Groot, 2006], especially on students with learning disabilities (Savan, 1999) or poor spelling skills (Scheree et al., 2000)

BITCOIN has been alarming people for years because of the amount of electricity needed to mint new virtual coinage

When I speak about web applications across the blockchain landscape, I purposely exclude exchange platforms like Kraken, Bitfinex or similar. These exchanges are conventional, centralized web services (which store your account credentials for example). Examples of decentralized services would be Chronobank, Waves Platform or BitBay’s web wallet.