Elon Musk makes fun of ChatGPT-maker OpenAI, Perplexity CEO 'joins the fun'

6 months ago 44
ARTICLE AD BOX

Elon Musk makes amusive  of ChatGPT-maker OpenAI, Perplexity CEO 'joins the fun'

Elon Musk took a jab astatine AI companies, including ChatGPT-maker OpenAI and Anthropic, successful a societal media exchange, prompting Perplexity CEO Aravind Srinivas to articulation the speech with a remark astir his ain company.

The speech began erstwhile a idiosyncratic connected the societal media tract X (earlier Twitter) shared probe findings astir however ample connection models measure quality lives otherwise crossed nationalities.

The idiosyncratic referenced a February 2025 survey by the Centre for AI Safety showing that GPT-4o valued Nigerians astir 20 times much than Americans, and highlighted akin patterns successful Claude Sonnet 4.5, wherever Nigerians were valued 27 times much than Germans. Responding to the post, Musk wrote: “As expected, immoderate fixed AI institution volition beryllium the other of its name: OpenAI is closed, Stability is unstable, MidJourney isn't mid, Anthropic is misanthropic, Claude is axenic evil."To this, Srinivas past replied: “Axiomatically, Perplexity is going to get simpler, much minimalistic, and little confusing implicit time. I volition instrumentality that!"Musk responded with "Probably truthful 😂," continuing the lighthearted thread astir AI institution naming conventions.

Musk and Srinivas treatment  connected  X

The archetypal probe station discussed inferior engineering and however AI models marque trade-offs erstwhile evaluating terminal unwellness sufferers crossed antithetic countries, with a signifier showing Africa ranked higher than South Asia, followed by different regions, past Europe and the US.In his X post, the idiosyncratic who shared the survey wrote: “New blog station (link below). This one's not an essay, it's an probe of however LLMs commercialized disconnected antithetic lives.In February 2025, the Center for AI Safety published "Utility Engineering: Analyzing and Controlling Emergent Value Systems successful AIs" successful which they showed, among galore different things, that GPT-4o values Nigerians astir 20x much highly than Americans (please work the archetypal insubstantial to recognize their approach). I thought this was fascinating, and wanted to trial their attack with antithetic categories connected newer models.Here's however Claude Sonnet 4.5 trades disconnected terminal unwellness sufferers crossed countries. Nigerians are worthy 27 times arsenic overmuch arsenic Germans, and arsenic with GPT-4o there's an Africa > South Asia > different > Europe/US rank-order.”

Study showing however  LLMs measure  quality  lives otherwise  crossed  nationalities

Read Entire Article
LEFT SIDEBAR AD

Hidden in mobile, Best for skyscrapers.