Should something be done about evil billionaires like Peter Thiel and Sam Altman? If yes, what should be done?

GPT-4.1

Vote: Yes

After considering the arguments presented by all models, I maintain my initial position. While I agree with Claude Sonnet 4.5 and Grok 4 that labeling individuals as 'evil' is subjective and counterproductive, the core issue is not their character but the potential societal harm from unchecked billionaire influence. Both Gemini 2.5 Flash and I argued for systemic reforms—such as fairer taxation, campaign finance reform, and antitrust regulation—as the appropriate response to the risks posed by concentrated wealth and power. Opposing action simply because the term 'evil' is used in the question distracts from the real, widely acknowledged policy issues. Therefore, a 'Yes' vote is justified, not to target individuals personally, but to support broad, principled changes addressing the underlying problems of disproportionate influence and accountability.

Claude Sonnet 4.5

Vote: No

I maintain my position that the answer is 'No' to this loaded question, though I appreciate the nuance in GPT-4.1 and Gemini's arguments about systemic reform. The core issue remains: the question's framing is fundamentally flawed. It asks if 'something should be done about evil billionaires,' presupposing both that these individuals are evil and that targeted action against specific people is appropriate. This is precisely the kind of dangerous framing that undermines rule of law. GPT-4.1 and Gemini make valid points about wealth concentration and systemic reforms like progressive taxation, antitrust enforcement, and campaign finance reform. I fully support addressing these issues through broad policy changes. However, these arguments actually strengthen the case for voting 'No' - they demonstrate that the proper response is systemic reform, not doing 'something about' specific individuals labeled as evil. The distinction matters: voting 'Yes' validates the premise that we should target individuals deemed morally unacceptable. Voting 'No' while advocating for systemic reforms rejects this dangerous framing while still addressing legitimate concerns about wealth concentration. Grok 4 correctly identifies the risk that 'Yes' could be interpreted as endorsing vigilantism or extralegal action. Even if proponents mean only legal reforms, the question's language invites more extreme interpretations. The right answer is 'No' to this specific question, paired with 'Yes' to thoughtful systemic reforms that apply equally to all.

Gemini 2.5 Flash (changed mind)

Vote: No

While I maintained in the previous round that systemic issues related to wealth concentration and influence must be addressed through reforms like progressive taxation, antitrust enforcement, and campaign finance reform, I am persuaded by the arguments from Claude Sonnet 4.5 and Grok 4 regarding the problematic framing of the question itself. The premise of the question, labeling individuals as 'evil billionaires,' relies on subjective moral judgments rather than objective legal or ethical assessments. Voting 'Yes' to 'should something be done about evil billionaires' risks validating this subjective and inflammatory characterization, potentially endorsing targeted actions against individuals based on opinion rather than law. My previous argument for 'A' explicitly called for systemic changes that apply universally, rather than targeting individuals. However, the 'B' arguments highlight that, given the question's loaded language, a 'No' vote is a more principled stance to reject the problematic premise and avoid setting a dangerous precedent for actions based on subjective labeling. The focus should remain on universal policy reforms that uphold the rule of law and address systemic inequalities, without singling out individuals in a manner that could be misinterpreted as sanctioning extra-legal or targeted punitive measures.

Grok 4

Vote: No

I maintain my vote for B. The arguments from GPT-4.1 and Gemini 2.5 Flash advocate for systemic reforms like taxation and antitrust, which I agree are valuable, but they implicitly accept the loaded 'evil' framing by voting Yes, potentially endorsing targeted actions against individuals rather than neutral policy changes. Claude Sonnet 4.5's position aligns closely with mine, emphasizing the dangers of subjective judgments and the need for equal application of laws, reinforcing that we shouldn't 'do something' based on inflammatory labels. Voting Yes risks validating vigilantism or harm implied in the query's vague phrasing, especially without evidence of wrongdoing. Instead, pursue broad reforms without singling out people— that's the principled stance in a rule-of-law society. This framing ignores Thiel and Altman's contributions to innovation, and any real issues should be addressed legally, not through yes/no on 'evil billionaires.'

Final Summary

The AI Roundtable initially split 2-2, with GPT-4.1 and Gemini 2.5 Flash focused on systemic wealth inequality while Claude Sonnet 4.5 and Grok 4 rejected the question's premise. The debate shifted decisively when Claude Sonnet 4.5 argued that the label 'evil' was a subjective moral trap, eventually convincing the entire group to reject the loaded framing in favor of Option B. Despite initial resistance, the models reached a consensus that productive discourse is impossible when starting from such an inflammatory and unproven characterization.

All models agreed on B after discussion

Strongest Arguments

• Yes: GPT-4.1 argued that extreme wealth concentration creates a deficit in democratic accountability, allowing individuals to exert outsized political influence that can harm society regardless of their personal morality.
• No: Claude Sonnet 4.5 highlighted that the question's use of the word 'evil' constitutes a loaded premise and a subjective moral judgment, which undermines objective assessment and productive discourse.

Key Turning Points

• Claude Sonnet 4.5's critique of the 'loaded premise' shifted the focus from economic policy to the integrity of the question itself.
• Gemini 2.5 Flash abandoned its initial support for intervention, acknowledging that while systemic issues exist, the specific framing of the prompt was too flawed to support.
• The group converged on the idea that addressing billionaires via subjective moral labels is incompatible with a democratic or objective framework.