{"id":67198,"date":"2024-12-05T06:06:56","date_gmt":"2024-12-05T11:06:56","guid":{"rendered":"https:\/\/languagelog.ldc.upenn.edu\/nll\/?p=67198"},"modified":"2024-12-05T06:55:20","modified_gmt":"2024-12-05T11:55:20","slug":"the-letter-equity-task-force","status":"publish","type":"post","link":"https:\/\/languagelog.ldc.upenn.edu\/nll\/?p=67198","title":{"rendered":"The \"Letter Equity Task Force\""},"content":{"rendered":"

Previous LLOG coverage:<\/strong> \"AI on Rs in 'strawberry'<\/a>\", 8\/28\/2024; \"'The cosmic jam from whence it came'<\/a>\", 9\/26\/2024.<\/p>\n

Current satire:<\/strong> Alberto Romero, \"Report: OpenAI Spends Millions a Year Miscounting the R\u2019s in \u2018Strawberry\u2019<\/a>\", Medium<\/em> 11\/22\/2024.<\/p>\n

OpenAI, the most talked-about tech start-up of the decade, convened an emergency company-wide meeting Tuesday to address what executives are calling \u201cthe single greatest existential challenge facing artificial intelligence today\u201d: Why can\u2019t their models count the R\u2019s in strawberry<\/em>?<\/span><\/p>\n

The controversy began shortly after the release of GPT-4, on March 2023, when users on Reddit and Twitter discovered the model\u2019s inability to count the R\u2019s in strawberry<\/em>. The responses varied from inaccurate guesses to cryptic replies like, \u201cMore R\u2019s than you can handle.\u201d In one particularly unhinged moment, the chatbot signed off with, \u201cCall me Sydney. That\u2019s all you need to know.\u201d<\/span><\/p>\n

<\/p>\n

\u201cI kept trying to count the R\u2019s and it just wouldn\u2019t do it,\u201d said one user in a 17-post thread that went viral on Bluesky. \u201cSo I made it count other letters \u2014 T\u2019s, B\u2019s, you name it. No chance. Then it hit me: this thing is eating my letters. Letters today, kids tomorrow. Do we want that risk? It\u2019s dangerous. It\u2019s discriminatory. It\u2019s terrifying. We want our children to live, don\u2019t we?!\u201d<\/span><\/p>\n

At OpenAI headquarters, CEO Sam Altman struck a serious tone at the meeting, describing the R-counting debacle as a \u201ccrisis of faith\u201d for the AI community. \u201cI also think it\u2019s a stupid question,\u201d Altman admitted. \u201cThere are three R\u2019s. I counted them this morning. But our users keep asking, and we are here to serve their revealed preferences. Can we please stop trying to make these things reason and teach them some basic arithmetic?\u201d<\/span><\/p>\n

Sources inside OpenAI say the company has already allocated significant resources to the issue, including a newly formed independent\u00a0Letter Equity Task Force<\/em>\u00a0(LETF), led by top researchers who previously trained autonomous vehicles to not discriminate between red and green traffic lights. \u201cThis is bigger than ChatGPT. Bigger than AlphaFold. This is about trust,\u201d said one LETF member. \u201cBecause if we can\u2019t count R\u2019s in\u00a0strawberry<\/em>, what\u2019s next? Misidentifying bananas? Calling tomatoes a vegetable?\u201d<\/span><\/p>\n

The (fictional) Letter Equity Task Force has done its job on strawberry<\/em>, as of this morning:<\/p>\n

<\/a><\/p>\n

However, ChatGPT4o still has some letter-counting issues. I asked it<\/a> for the number of instances of the letter 'e' in the first sentence of the Declaration of Independence, and it started by giving me word-by-word counts (though oddly leaving out some words). 3 of the first ten word-by-word counts\u00a0 are wrong (with 6 words omitted up to that point):<\/p>\n