First Page Design
Site Theme
One year ago today, on November 30, 2022, OpenAI launched ChatGPT. It’s rare for a single product to have as much impact on the tech industry as ChatGPT in just one year.
Imagine a computer that can talk to you. Nothing new, right? Those have been around since the 1960s. But ChatGPT, the application that first brought large language models (LLMs) to a wide audience, felt different. It could compose poetry, seemingly understand the context of your questions and your conversation, and help you solve problems. Within a few months, it became the fastest-growing consumer application of all time. And it created a frenzy in the tech world.
Over the past 365 days, ChatGPT has amplified the public’s belief in AI, captured the imagination, attracted criticism, and stoked existential angst. It has emboldened and reoriented Microsoft, made Google dance, stoked fears about AGI’s global dominance, captivated world leaders, incited attempts at government regulation, helped load words into dictionaries, encouraged lectures and imitators, sparked a crisis for educators, hypercharged and automated smears, lawyers embarrassed by hallucinations, lawsuits for educational data and much more.
For some, ChatGPT has also proven to be a useful tool, speeding up some software developers’ programming responsibilities, helping writers with writing and editing responsibilities, and offering quick, ad-free advice using cookie banners. some social networking sites have fractured and our old oracle of dubious internet knowledge, Google, has fallen into terrible disrepair.
“Imagine if every human being could automate the tedious, repetitive information tasks in their lives, without needing to first get a computer science degree,” AI researcher Simon Willison told Ars in an interview about ChatGPT’s impact. “I’m seeing glimpses that LLMs might help make a huge step in that direction.”
Despite all the impact, none of this was planned before the release of ChatGPT. In fact, it started as a generational manifestation. According to a fair article on the creation of ChatGPT from the MIT Technology Review published in March (which could well be 3,000 years old), OpenAI saw it as a review of the research, a way to gather feedback on its flaws. open to the public. “We didn’t need to offer it as a basic breakthrough,” OpenAI scientist Liam Fedus told this publication.
ChatGPT relies on a generation that tries to wait for the next likely maximum token, or word fragment, in a sequence. In 2019, after the release of GPT-2, someone learned that this generation could become a chatbot if The Giant Language Style won the transcription of a verbal exchange that had to end. With this method, every spark in the language style includes the full story of the verbal exchange. In the GPT-3 era, OpenAI included a chat-like implementation of this feature in the developer test domain (called “Playground”) on its website, but it’s not publicly available. With ChatGPT, OpenAI took this concept, simplified it by refining an edit of GPT-3 in chat transcripts, and released it. for the audience to play.
From a tech point of view, ChatGPT didn’t initially strike many as being especially novel, among those paying attention to AI. Chatbots had existed for some time. GPT-3, the AI model that led to ChatGPT, had been out since 2020. In September 2022, a compelling AI chat site called Character.ai launched, and we used it to interview “Linux.” And just two weeks before ChatGPT, Meta had been publicly shamed for launching a public preview of Galactica, an AI model that purported to write science literature but could also hallucinate racist nonsense.
But compared to Galactica, OpenAI technologized things from a different angle with ChatGPT. From the beginning, the company adopted a modest and cautious strategy that allowed the experiment to continue even despite rigorous public testing. From the start, ChatGPT refused. To answer some incendiary questions, and while astute users looking for trouble on social media circumvented any and all limitations, OpenAI erected new barriers to keep ChatGPT online. These limitations frustrated many people, who hated to hold hands artificially, but avoided media outbursts that might otherwise have killed the project.
OpenAI CEO Sam Altman has also been open about ChatGPT’s limitations. Shortly after its launch, he tweeted, “ChatGPT is incredibly limited, but smart enough in some spaces to create a misleading impression of grandeur. “There’s something important going on right now. It’s a glimmer of progress; We have a lot of work to do on robustness and veracity. “Later, he wrote, “He knows a lot, but the danger is that he will have confidence in himself and make the most of it. “of time. “
But ChatGPT never stopped. OpenAI tests new features with a wide audience, who occasionally use them for free, although a paid tier for priority access (ChatGPT Plus) was added in February. With the release of GPT-4 in March 2023, ChatGPT got a dramatic upgrade, removing since then, OpenAI has added voice conversations, symbol generation, and interpretation to ChatGPT. To this day, ChatGPT’s GPT-4 remains widely regarded as the leader among AI language models, even as giants like Google rush to catch up with PalM and Gemini.
Join the Ars Orbital Transmission email to receive weekly updates in your inbox. Sign up →