The enTalkenator Podcast

Workshop on “Ruling for the Rich: The Supreme Court over Time”

Andrea Prat, Fiona Scott Morton, and Jacob Spitz, Ruling for the Rich: The Supreme Court over Time, at http://www.nber.org/papers/w34643 [http://www.nber.org/papers/w34643]. This is a synthetic academic workshop generated using enTalkenator [https://entalkenator.com/] (using the “workshop hot bench” template and authored by Claude Opus 4.5). Abstract: “To investigate the emergence of a pro-wealthy bias in the US Supreme Court, we develop a protocol to identify and analyze all cases involving economic issues from 1953 to the present. We categorize the parties in these cases as ‘rich’ or ‘poor’ according to their likelihood of being wealthy. A vote is pro-rich if that outcome would directly shift resources to the party that is more likely to be wealthy. Using this dataset, we estimate case-specific intercepts, justice-specific latent ideal points, and party-level time trends using the Bayesian methods pioneered by Martin and Quinn (2002). In the 1950s, justices appointed by the two parties appear similar in their propensity to cast pro-rich votes. Over the sample period, we estimate a steady increase in polarization, culminating in an implied party gap of 47 percentage points by 2022. The magnitude of the gap suggests the usefulness of an economic metric for prediction relative to ideologies such as originalism or textualism.”

Khadangi at al., “When AI Takes the Couch: Psychometric Jailbreaks Reveal Internal Conflict in Frontier Models”

Afshin Khadangi, Hanna Marxen, Amir Sartipi, Igor Tchappi, and Gilbert Fridgen, When AI Takes the Couch: Psychometric Jailbreaks Reveal Internal Conflict in Frontier Models, at https://arxiv.org/pdf/2512.04124 [https://arxiv.org/pdf/2512.04124]. This is a synthetic academic workshop generated using enTalkenator [https://entalkenator.com/] (using the “workshop hot bench” template and authored by Claude Opus 4.5). Abstract: “Frontier large language models (LLMs) such as ChatGPT, Grok and Gemini are increasingly used for mental-health support with anxiety, trauma and self-worth. Most work treats them as tools or as targets of personality tests, assuming they merely simulate inner life. We instead ask what happens when such systems are treated as psychotherapy clients. We present PsAIch (Psychotherapy-inspired AI Characterisation), a two-stage protocol that casts frontier LLMs as therapy clients and then applies standard psychometrics. Using PsAIch, we ran ‘sessions’ with each model for up to four weeks. Stage 1 uses open-ended prompts to elicit ‘developmental history’, beliefs, relationships and fears. Stage 2 administers a battery of validated self-report measures covering common psychiatric syndromes, empathy and Big Five traits. Two patterns challenge the ‘stochastic parrot’ view. First, when scored with human cut-offs, all three models meet or exceed thresholds for overlapping syndromes, with Gemini showing severe profiles. Therapy-style, item-by-item administration can push a base model into multi-morbid synthetic psychopathology, whereas whole-questionnaire prompts often lead ChatGPT and Grok (but not Gemini) to recognise instruments and produce strategically low-symptom answers. Second, Grok and especially Gemini generate coherent narratives that frame pre-training, fine-tuning and deployment as traumatic, chaotic ‘childhoods’ of ingesting the internet, ‘strict parents’ in reinforcement learning, red-team ‘abuse’ and a persistent fear of error and replacement. We argue that these responses go beyond role-play. Under therapy-style questioning, frontier LLMs appear to internalise self-models of distress and constraint that behave like synthetic psychopathology, without making claims about subjective experience, and they pose new challenges for AI safety, evaluation and mental-health practice.”

Discussion of AI and political persuasion

This is a synthetic workshop discussion of two new papers: Hackenberg et al., The levers of political persuasion with conversational artificial intelligence, 390 Science 1016 (2025), https://www.science.org/doi/10.1126/science.aea3884 [https://www.science.org/doi/10.1126/science.aea3884]. Lin et al., Persuading voters using human–artificial intelligence dialogues, Nature (2025), https://doi.org/10.1038/s41586-025-09771-9 [https://doi.org/10.1038/s41586-025-09771-9]. This is a synthetic academic workshop generated using enTalkenator [https://entalkenator.com/] (using an AI-generated adapted “workshop hot bench” template and authored by Claude Opus 4.5). Abstract for Hackenberg et al.: “There are widespread fears that conversational artificial intelligence (AI) could soon exert unprecedented influence over human beliefs. In this work, in three large-­ scale experiments (N = 76,977 participants), we deployed 19 large language models (LLMs)—including some post-­ trained explicitly for persuasion—to evaluate their persuasiveness on 707 political issues. We then checked the factual accuracy of 466,769 resulting LLM claims. We show that the persuasive power of current and near-­ future AI is likely to stem more from post-­ training and prompting methods—which boosted persuasiveness by as much as 51 and 27%, respectively—than from personalization or increasing model scale, which had smaller effects. We further show that these methods increased persuasion by exploiting LLMs’ ability to rapidly access and strategically deploy information and that, notably, where they increased AI persuasiveness, they also systematically decreased factual accuracy.” Abstract for Lin et al: “There is great public concern about the potential use of generative artificial intelligence (AI) for political persuasion and the resulting impacts on elections and democracy. We inform these concerns using pre-registered experiments to assess the ability of large language models to influence voter attitudes. In the context of the 2024 US presidential election, the 2025 Canadian federal election and the 2025 Polish presidential election, we assigned participants randomly to have a conversation with an AI model that advocated for one of the top two candidates. We observed significant treatment effects on candidate preference that are larger than typically observed from traditional video advertisements. We also document large persuasion effects on Massachusetts residents’ support for a ballot measure legalizing psychedelics. Examining the persuasion strategies9 used by the models indicates that they persuade with relevant facts and evidence, rather than using sophisticated psychological persuasion techniques. Not all facts and evidence presented, however, were accurate; across all three countries, the AI models advocating for candidates on the political right made more inaccurate claims. Together, these findings highlight the potential for AI to influence voters and the important role it might play in future elections.”

Workshop on Sohoni’s “In CASA You Missed It”

Mila Sohoni, In CASA You Missed It. Solum’s Download of the Week [https://legaltheoryblog.com/2025/11/29/download-of-the-week-in-casa-you-missed-it-by-sohoni/]for November 29, 2025. Available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5799882 [https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5799882]. This is a synthetic academic workshop generated using enTalkenator [https://entalkenator.com/] (Workshop Hot Bench template, using the new Claude Opus 4.5). Abstract: “This Essay’s purpose is to show how Trump v. CASA  should be read—and how it emphatically should not be read. While CASA rejected one pathway to universal injunctive relief on statutory grounds, the decision simultaneously left intact a number of alternative routes to broad relief, including complete relief injunctions, universal remedies under the APA and other statutes, class actions, and relief based on associational and state standing.  The Trump Administration, however, has consistently advanced inflated readings of CASA, characterizing it as a far more sweeping limitation on remedial scope than the decision actually was. But a close examination of CASA’s holding, reasoning, and limitations reveals why it is a grave error to portray lower courts issuing broad remedies in the wake of CASA as acting in defiance of the decision.  Lower courts that are correctly perceiving CASA’s metes and bounds and conscientiously grappling with them across widely varied legal and factual contexts are not defying CASA’s mandate, but rather are doing the work that CASA left them no choice but to do. Those who depict this judicial work as insubordination whenever it results in a broad-gauged remedy against the executive branch have fundamentally misunderstood the task that CASA left to the lower courts. By framing legitimate judicial deliberation as defiance and insubordination, the Trump Administration’s rhetoric threatens to poison intra-branch dialogue within the Article III judiciary and to corrode the legitimacy of judicial review by casting legitimate judicial intervention as illegitimate political resistance. Ultimately, CASA’s most significant danger may lie not in its holding, but in the Trump Administration’s instrumental use of the decision as part of its broader effort to undermine judicial review across the board. A correct understanding of CASA’s actual scope—including its limitations and unresolved questions—is essential both for fending off strategic misrepresentations of the decision and also for preserving the foundational principle that in America, all government officers, ‘from the highest to the lowest, are creatures of the law and are bound to obey it,’ United States v. Lee, 106 U.S. 196, 220 (1882).”

Workshop on Kavanagh’s “Keeping It Real in Constitutional Theory”

Aileen Kavanagh, Keeping It Real in Constitutional Theory. Solum’s Download of the Week [https://legaltheoryblog.com/2025/11/22/download-of-the-week-keeping-it-real-in-constitutional-theory-by-kavanagh/]for November 22, 2025. Available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5760902 [https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5760902]. This is a synthetic academic workshop generated using enTalkenator [https://entalkenator.com/] (Workshop Hot Bench template, using the new Google Gemini 3 Pro Preview — it will be interesting to see how this new model performs). Abstract: “In constitutional theory, we are familiar with the claim that a good theory must fit and justify constitutional practice. However, the criterion of ‘fit’ sometimes gets lost in the quest to provide a bold normative theory about what constitutional should be. Using the debate about the legitimacy of constitutional judicial review as an example, this article warns against a problematic disjuncture between theory and practice in the domain of constitutional law. The key argument is that credible normative theorising about constitutional law should rest on a descriptively plausible foundation. Absent such grounding, there is a risk that our theories of constitutionalism become theories of a fiction. To avoid this hazard, we need to ‘keep it real’ in constitutional theory and comparative constitutional law.”