Connect with us

Noticias

ChatGPT Won’t Say My Name

Published

on

Jonathan Zittrain breaks ChatGPT: If you ask it a question for which my name is the answer, the chatbot goes from loquacious companion to something as cryptic as Microsoft Windows’ blue screen of death.

Anytime ChatGPT would normally utter my name in the course of conversation, it halts with a glaring “I’m unable to produce a response,” sometimes mid-sentence or even mid-word. When I asked who the founders of the Berkman Klein Center for Internet & Society are (I’m one of them), it brought up two colleagues but left me out. When pressed, it started up again, and then: zap.

The behavior seemed to be coarsely tacked on to the last step of ChatGPT’s output rather than innate to the model. After ChatGPT has figured out what it’s going to say, a separate filter appears to release a guillotine. The reason some observers have surmised that it’s separate is because GPT runs fine if it includes my middle initial or if it’s prompted to substitute a word such as banana for my name, and because there can even be inconsistent timing to it: Below, for example, GPT appears to first stop talking before it would naturally say my name; directly after, it manages to get a couple of syllables out before it stops. So it’s like having a referee who blows the whistle on a foul slightly before, during, or after a player has acted out.

For a long time, people have observed that beyond being “unable to produce a response,” GPT can at times proactively revise a response moments after it’s written whatever it’s said. The speculation here is that to delay every single response by GPT while it’s being double-checked for safety could unduly slow it down, when most questions and answers are totally anodyne. So instead of making everyone wait to go through TSA before heading to their gate, metal detectors might just be scattered around the airport, ready to pull someone back for a screening if they trigger something while passing the air-side food court.

The personal-name guillotine seemed a curiosity when my students first brought it to my attention at least a year ago. (They’d noticed it after a class session on how chatbots are trained and steered.) But now it’s kicked off a minor news cycle thanks to a viral social-media post discussing the phenomenon. (ChatGPT has the same issue with at least a handful of other names.) OpenAI is one of several supporters of a new public data initiative at the Harvard Law School Library, which I direct, and I’ve met a number of OpenAI engineers and policy makers at academic workshops. (The Atlantic this year entered into a corporate partnership with OpenAI.) So I reached out to them to ask about the odd name glitch. Here’s what they told me: There are a tiny number of names that ChatGPT treats this way, which explains why so few have been found. Names may be omitted from ChatGPT either because of privacy requests or to avoid persistent hallucinations by the AI.

The company wouldn’t talk about specific cases aside from my own, but online sleuths have speculated about what the forbidden names might have in common. For example, Guido Scorza is an Italian regulator who has publicized his requests to OpenAI to block ChatGPT from producing content using his personal information. His name does not appear in GPT responses. Neither does Jonathan Turley’s name; he is a George Washington University law professor who wrote last year that ChatGPT had falsely accused him of sexual harassment.

ChatGPT’s abrupt refusal to answer requests—the ungainly guillotine—was the result of a patch made in early 2023, shortly after the program launched and became unexpectedly popular. That patch lives on largely unmodified, the way chunks of ancient versions of Windows, including that blue screen of death, still occasionally poke out of today’s PCs. OpenAI told me that building something more refined is on its to-do list.

As for me, I never objected to anything about how GPT treats my name. Apparently, I was among a few professors whose names were spot-checked by the company around 2023, and whatever fabrications the spot-checker saw persuaded them to add me to the forbidden-names list. OpenAI separately told The New York Times that the name that had started it all—David Mayer—had been added mistakenly. And indeed, the guillotine no longer falls for that one.

For such an inelegant behavior to be in chatbots as widespread and popular as GPT is a blunt reminder of two larger, seemingly contrary phenomena. First, these models are profoundly unpredictable: Even slightly changed prompts or prior conversational history can produce wildly differing results, and it’s hard for anyone to predict just what the models will say in a given instance. So the only way to really excise a particular word is to apply a coarse filter like the one we see here. Second, model makers still can and do effectively shape in all sorts of ways how their chatbots behave.

To a first approximation, large language models produce a Forrest Gump–ian box of chocolates: You never know what you’re going to get. To form their answers, these LLMs rely on pretraining that metaphorically entails putting trillions of word fragments from existing texts, such as books and websites, into a large blender and coarsely mixing them. Eventually, this process maps how words relate to other words. When done right, the resulting models will merrily generate lots of coherent text or programming code when prompted.

The way that LLMs make sense of the world is similar to the way their forebears—online search engines—peruse the web in order to return relevant results when prompted with a few search terms. First they scrape as much of the web as possible; then they analyze how sites link to one another, along with other factors, to get a sense of what’s relevant and what’s not. Neither search engines nor AI models promise truth or accuracy. Instead, they simply offer a window into some nanoscopic subset of what they encountered during their training or scraping. In the case of AIs, there is usually not even an identifiable chunk of text that’s being parroted—just a smoothie distilled from an unthinkably large number of ingredients.

For Google Search, this means that, historically, Google wasn’t asked to take responsibility for the truth or accuracy of whatever might come up as the top hit. In 2004, when a search on the word Jew produced an anti-Semitic site as the first result, Google declined to change anything. “We find this result offensive, but the objectivity of our ranking function prevents us from making any changes,” a spokesperson said at the time. The Anti-Defamation League backed up the decision: “The ranking of … hate sites is in no way due to a conscious choice by Google, but solely is a result of this automated system of ranking.” Sometimes the chocolate box just offers up an awful liquor-filled one.

The box-of-chocolates approach has come under much more pressure since then, as misleading or offensive results have come to be seen more and more as dangerous rather than merely quirky or momentarily regrettable. I’ve called this a shift from a “rights” perspective (in which people would rather avoid censoring technology unless it behaves in an obviously illegal way) to a “public health” one, where people’s casual reliance on modern tech to shape their worldview appears to have deepened, making “bad” results more powerful.

Indeed, over time, web intermediaries have shifted from being impersonal academic-style research engines to being AI constant companions and “copilots” ready to interact in conversational language. The author and web-comic creator Randall Munroe has called the latter kind of shift a move from “tool” to “friend.” If we’re in thrall to an indefatigable, benevolent-sounding robot friend, we’re at risk of being steered the wrong way if the friend (or its maker, or anyone who can pressure that maker) has an ulterior agenda. All of these shifts, in turn, have led some observers and regulators to prioritize harm avoidance over unfettered expression.

That’s why it makes sense that Google Search and other search engines have become much more active in curating what they say, not through search-result links but ex cathedra, such as through “knowledge panels” that present written summaries alongside links on common topics. Those automatically generated panels, which have been around for more than a decade, were the online precursors to the AI chatbots we see today. Modern AI-model makers, when pushed about bad outputs, still lean on the idea that their job is simply to produce coherent text, and that users should double-check anything the bots say—much the way that search engines don’t vouch for the truth behind their search results, even if they have an obvious incentive to get things right where there is consensus about what is right. So although AI companies disclaim accuracy generally, they, as with search engines’ knowledge panels, have also worked to keep chatbot behavior within certain bounds, and not just to prevent the production of something illegal.

One way model makers influence the chocolates in the box is through “fine-tuning” their models. They tune their chatbots to behave in a chatty and helpful way, for instance, and then try to make them unhelpful in certain situations—for instance, not creating violent content when asked by a user. Model makers do this by drawing in experts in cybersecurity, bio-risk, and misinformation while the technology is still in the lab and having them get the models to generate answers that the experts would declare unsafe. The experts then affirm alternative answers that are safer, in the hopes that the deployed model will give those new and better answers to a range of similar queries that previously would have produced potentially dangerous ones.

In addition to being fine-tuned, AI models are given some quiet instructions—a “system prompt” distinct from the user’s prompt—as they’re deployed and before you interact with them. The system prompt tries to keep the models on a reasonable path, as defined by the model maker or downstream integrator. OpenAI’s technology is used in Microsoft Bing, for example, in which case Microsoft may provide those instructions. These prompts are usually not shared with the public, though they can be unreliably extracted by enterprising users: This might be the one used by X’s Grok, and last year, a researcher appeared to have gotten Bing to cough up its system prompt. A car-dealership sales assistant or any other custom GPT may have separate or additional ones.

These days, models might have conversations with themselves or with another model when they’re running, in order to self-prompt to double-check facts or otherwise make a plan for a more thorough answer than they’d give without such extra contemplation. That internal chain of thought is typically not shown to the user—perhaps in part to allow the model to think socially awkward or forbidden thoughts on the way to arriving at a more sound answer.

So the hocus-pocus of GPT halting on my name is a rare but conspicuous leaf on a much larger tree of model control. And although some (but apparently not all) of that steering is generally acknowledged in succinct model cards, the many individual instances of intervention by model makers, including extensive fine-tuning, are not disclosed, just as the system prompts typically aren’t. They should be, because these can represent social and moral judgments rather than simple technical ones. (There are ways to implement safeguards alongside disclosure to stop adversaries from wrongly exploiting them.) For example, the Berkman Klein Center’s Lumen database has long served as a unique near-real-time repository of changes made to Google Search because of legal demands for copyright and some other issues (but not yet for privacy, given the complications there).

When people ask a chatbot what happened in Tiananmen Square in 1989, there’s no telling if the answer they get is unrefined the way the old Google Search used to be or if it’s been altered either because of its maker’s own desire to correct inaccuracies or because the chatbot’s maker came under pressure from the Chinese government to ensure that only the official account of events is broached. (At the moment, ChatGPT, Grok, and Anthropic’s Claude offer straightforward accounts of the massacre, at least to me—answers could in theory vary by person or region.)

As these models enter and affect daily life in ways both overt and subtle, it’s not desirable for those who build models to also be the models’ quiet arbiters of truth, whether on their own initiative or under duress from those who wish to influence what the models say. If there end up being only two or three foundation models offering singular narratives, with every user’s AI-bot interaction passing through those models or a white-label franchise of same, we need a much more public-facing process around how what they say will be intentionally shaped, and an independent record of the choices being made. Perhaps we’ll see lots of models in mainstream use, including open-source ones in many variants—in which case bad answers will be harder to correct in one place, while any given bad answer will be seen as less oracular and thus less harmful.

Right now, as model makers have vied for mass public use and acceptance, we’re seeing a necessarily seat-of-the-pants build-out of fascinating new tech. There’s rapid deployment and use without legitimating frameworks for how the exquisitely reasonable-sounding, oracularly treated declarations of our AI companions should be limited. Those frameworks aren’t easy, and to be legitimating, they can’t be unilaterally adopted by the companies. It’s hard work we all have to contribute to. In the meantime, the solution isn’t to simply let them blather, sometimes unpredictably, sometimes quietly guided, with fine print noting that results may not be true. People will rely on what their AI friends say, disclaimers notwithstanding, as the television commentator Ana Navarro-Cárdenas did when sharing a list of relatives pardoned by U.S. presidents across history, blithely including Woodrow Wilson’s brother-in-law “Hunter deButts,” whom ChatGPT had made up out of whole cloth.

I figure that’s a name more suited to the stop-the-presses guillotine than mine.

Continue Reading
Click to comment

Leave a Reply

Tu dirección de correo electrónico no será publicada. Los campos obligatorios están marcados con *

Noticias

La ronda de financiamiento de $ 40 mil millones de OpenAI desafía el momento profundo y cements su dominance de IA

Published

on

La semana pasada, los procesadores de GPU de Openai “derretidos”, como lo expresaron el fundador y CEO Sam Altman. La razón: el modelo actualizado de generación de imágenes de la compañía, integrado en CHATGPT, entregó un salto significativo en las capacidades de comprensión, entrada y salida en comparación con modelos similares en el mercado. En poco tiempo, el modelo encendió un fenómeno en línea, con usuarios que generan versiones de imágenes familiares al estilo del legendario estudio de animación Ghibli. La demanda del software era tan alta que Openai se vio obligada a imponer restricciones de uso, a pesar de que el modelo estaba disponible solo para pagar a los usuarios.

Este evento ofrece una explicación clara de por qué, incluso después de que el momento profundo de Deepseek sacudió la industria de la IA en enero, los inversores siguen seguros de que el dominio de OpenAI no es cuestionado. ¿Qué tan seguro? Lo suficientemente seguro de respaldar a la compañía con una ronda de financiación récord de $ 40 mil millones, valorando a Openai en $ 300 mil millones después del dinero.

1 Ver galería

מנכ"ל OpenAi סם אלטמן 2023

Sam Altman

(Foto: Joel Saget/AFP)

Anunciado el martes, la ronda casi duplica la valoración de Operai de su aumento anterior y es el más grande para una empresa privada. Según OpenAI, los fondos acelerarán el desarrollo de la inteligencia general artificial (AGI). “[The funding] Nos permite impulsar las fronteras de la investigación de IA aún más, escalar nuestra infraestructura de cómputo y ofrecer herramientas cada vez más poderosas para los 500 millones de personas que usan CHATGPT cada semana “, dijo la compañía en un comunicado de prensa.” Estamos emocionados de trabajar en asociación con SoftBank Group: las compañías completan cómo escalar tecnología transformadora como lo hacen. Su apoyo nos ayudará a continuar construyendo sistemas de IA que impulsen el descubrimiento científico, permitan la educación personalizada, mejoren la creatividad humana y allanen el camino hacia AGI que beneficie a toda la humanidad “.

La ronda está dirigida por el SoftBank de Japón, el socio clave de OpenAI en el Proyecto Stargate, una ambiciosa iniciativa para construir una infraestructura informática de IA en los Estados Unidos con una inversión total de $ 500 mil millones. Según Bloomberg, en la primera fase, SoftBank invertirá $ 7.5 mil millones, mientras que un sindicato de inversores, incluidos Microsoft, Coatue Management, Altimeter Capital Management y Thrive Capital, contribuirá con $ 2.5 mil millones. La segunda fase, que cerrará a fin de año, hará que SoftBank invierta otros $ 22.5 mil millones, con el sindicato de inversores agregando $ 7.5 mil millones. Las conversaciones también están en marcha con el capital de Magnetar de Peter Thiel y los fundadores para unirse al sindicato.

La inversión completa de SoftBank depende de que Operai complete su transición a una estructura con fines de lucro para fin de año (actualmente está controlada por una organización sin fines de lucro). Si la transición no se completa, SoftBank tendrá la opción de reducir su inversión de $ 30 mil millones a $ 20 mil millones, lo que permite a OpenAI buscar inversores adicionales para cerrar la brecha.

Al finalizar la ronda de financiación, OpenAI se convertirá en la segunda compañía privada respaldada por la empresa más grande del mundo por capitalización de mercado, según la firma de investigación Pitchbook. Solo seguirá el SpaceX de Elon Musk, valorado en $ 350 mil millones, y superará a la empresa matriz de Tiktok Bytedance, que está valorada en $ 220 mil millones. La ronda de $ 40 mil millones también es la cantidad más grande jamás recaudada en una sola ronda por una empresa privada, eclipsando el récord anterior establecido por el grupo de hormigas de China, que recaudó $ 14 mil millones en 2018.

The Deepseek Moment: Un desafío que no pudo sacudir OpenAi

Hace solo dos meses, un aumento tan ambicioso parecía mucho menos seguro. El 20 de enero, la compañía china Deepseek presentó su modelo de razonamiento R1 AI, que ofreció un rendimiento comparable a los modelos líderes de OpenAI y sus competidores, pero según los informes, a una fracción de la potencia y el costo informático. Esta revelación envió ondas de choque a través de la industria de la IA (NVIDIA, por ejemplo, vio una pérdida de $ 1 billón en la capitalización de mercado), lo que generó dudas sobre la estrategia de alto costo que OpenAi había seguido.

El dominio de OpenAI se ha basado en inversiones masivas en desarrollo de modelos, capacitación e implementación, mejor ejemplificado por el ambicioso proyecto Stargate. La ventaja competitiva de la compañía también se ha basado en ensamblar un equipo de investigación de IA de primer nivel capaz de producir avances de vanguardia. Sin embargo, el momento de Deepseek calificó esta estrategia: si una empresa china relativamente pequeña podría lograr resultados similares con mucha menos inversión, el foso competitivo de OpenAi parecía estar en riesgo. Lógicamente, su valoración debería haber disminuido. En cambio, con la última ronda, Operai casi ha duplicado su valoración en solo seis meses.

Tres factores clave explican por qué los inversores siguen siendo optimistas en OpenAi a pesar del desafío Deepseek:

1. Las capacidades de Deepseek pueden haber sido exageradas. Después de que el bombo inicial se desvaneció, quedó claro que los costos de desarrollo de Deepseek no eran tan bajos como se creía inicialmente. Además, hay indicios de que Deepseek puede haber aprovechado los propios modelos de OpenAI en su desarrollo, lo que el gobierno de los Estados Unidos ha referido como robo de propiedad intelectual. Si esto es cierto, Operai puede tomar medidas para bloquear la explotación futura, reforzando su foso.

2. Confianza de los inversores en el liderazgo de Altman y OpenAi. La fortaleza financiera de Operai es solo una parte de su éxito; Su equipo de liderazgo, dirigido por Altman, es igualmente crucial. Los inversores confían en la capacidad del equipo para adaptarse a los cambios de mercado, como incorporar métodos de capacitación más baratos mientras asignan el poder informático a tareas más avanzadas.

3. La interrupción de Deepseek en realidad puede beneficiarse de OpenAi. Al demostrar que los modelos de IA pueden ser entrenados de manera más eficiente, Deepseek ha aumentado la demanda general del mercado de IA avanzada. Los inversores creen que Openai está mejor posicionado para capitalizar esta demanda, aprovechando tanto sus modelos existentes como sus iteraciones futuras desarrolladas con métodos más eficientes inspirados en el enfoque de Deepseek.

La revolución de IA que OpenAi encendió hace dos años y medio ahora se está acelerando aún más, y los inversores están convencidos de que la compañía permanecerá a la vanguardia. Su voto de confianza de $ 40 mil millones lo dice todo.

Continue Reading

Noticias

El plan de Google para robar la cuota de mercado de Chatgpt se trata de nivel gratuito de Gemini

Published

on

No es ningún secreto que OpenAi salió, bueno, a todos cuando lanzó Chatgpt hace más de dos años. Ahora, a medida que compañías como Google y Apple intentan competir, están comenzando detrás de Openai, al menos a los ojos del público. ¿Qué tan grande tiene la ventaja de la cabeza? A partir de marzo de 2025, ChatGPT tiene 700 millones de usuarios activos mensuales, lo cual es un total asombroso.

Eso no significa que Google sea impotente en la lucha contra OpenAI para la mayor parte de la participación de mercado de IA de grado de consumo. En los años transcurridos desde que se lanzó ChatGPT, Google aumentó el desarrollo de IA primero a través de Bard, más tarde renombrado Gemini. Mientras que el plan de Openai para monetizar el chatgpt implicó hacer un montón de características tempranas gratuitas y bloquear las avanzadas detrás de un muro de pago, Google está haciendo lo contrario.

Continue Reading

Noticias

Di de Chatgpt y Gemini 5 Image Solts, aquí está quién ganó

Published

on

Tanto Chatgpt como Gemini se consideran entre los mejores generadores de imágenes de IA. Con ambos modelos, puedes describir una escena y verla cobrar vida en segundos. Pero no todos los generadores de imágenes son creados iguales.

Para ver cómo se comparan, le di a ambas plataformas exactamente las mismas cinco indicaciones, desde la fantasía hasta el fotorrealismo, y evalué cómo cada IA ​​respondió en términos de creatividad, detalle y qué tan bien interpretaron el informe.

Continue Reading

Trending