Like most of you, we’ve been transfixed by the arrival of ChatGPT, the large language model developed by OpenAI that understands and responds to natural language input. Its ability to generate human-like responses to a wide range of questions and text prompts even conceals its shortcomings – for example, that its cutoff date for knowledge training was September 2021 (though it continues to incorporate new information).
What makes ChatGPT so remarkable is its ease of use. It responds to our opinions, ideas, and factual questions and engages in wide-ranging conversations . . . but truthfully, I’m shortchanging ChatGPT by summarizing it this way. It’s actually easier to let it speak for itself. When I asked it what it could do, here was its response:
As a language model, I can perform a wide range of natural language processing tasks, including:
- Answering questions: I can provide answers to factual questions, as well as questions that require a more nuanced understanding of a topic.
- Translation: I can translate text from one language to another.
- Text summarization: I can summarize large bodies of text into a shorter, more condensed version.
- Text generation: I can generate human-like text in response to a prompt or topic.
- Sentiment analysis: I can analyze text to determine the overall sentiment or mood of the text.
- Conversational AI: I can engage in natural language conversations with users on a wide range of topics.
- Language modeling: I can generate coherent and grammatically correct sentences or paragraphs based on a given prompt or topic.
And then, when I asked if that was all – assuming there might be a small degree of humility from the new AI kid on the block – it continued:
- Text classification: I can classify text into different categories or topics.
- Image captioning: I can generate descriptions or captions for images.
- Knowledge retrieval: I can retrieve information from a large knowledge base and provide answers to complex questions.
- Chatbot development: I can be integrated into chatbot systems to provide conversational interfaces for various applications.
- Sentiment analysis: I can analyze the sentiment of a large volume of text, such as social media posts, to identify patterns and trends. (OpenAI, ChatGPT, February 20, 2023)
Helpfully, it even continued the numbered sequence so that I – as perhaps a flawed human – wouldn’t be confused by two numerical lists. But its self-generated list of capabilities is one thing – and something it will happily do over and over again (it is, after all, an AI-driven chatbot). Where it gets interesting is how it will evolve into the future and the impact of more powerful subsequent versions on society.
We’re Reliving the iPhone Moment With AI
If there’s a parallel to the moment we find ourselves in today with ChatGPT, it’s the introduction of Apple’s iPhone. It’s hard to forget that presentation (which was total tech theater) at the Macworld Expo on January 9, 2007, when Steve Jobs uttered words that would go down in history,
In 1984, we introduced the Macintosh. It didn’t just change Apple, it changed the whole computer industry. In 2001, we introduced the first iPod. And it didn’t just– it didn’t just change the way we all listen to music, it changed the entire music industry.
Today, we’re introducing three revolutionary products of this class. The first one is a widescreen iPod with touch controls. The second is a revolutionary mobile phone. And the third is a breakthrough internet communications device. So three things. A widescreen iPod with touch controls, a revolutionary mobile phone, and a breakthrough internet communications device . . .
An iPod, a phone, and an internet communicator. An iPod–a phone–Are you getting it . . . ?
These are not three separate devices. This is one device. And we are calling it iPhone. Today, Apple is going to reinvent the phone. And here it is. [Transcript]
And it was revolutionary. The iPhone was unique for its multi-touch display and onscreen keyboard, an accelerometer that let you switch back and forth between landscape and portrait mode, a camera, the introduction of apps (15 in all), and a new operating system (iPhoneOS).
Underestimating the Impact of ChatGPT
But while many saw the iPhone as a breakthrough device, others were underwhelmed by its limited capabilities. A 2007 Fast Company article listed many of the complaints, from cost to incompatibility with popular email clients, typing on a glass screen (Job’s classic response: “You’ll get used to it.”), and a camera that was inferior to Nokia’s high-tech camera on the N95. And we forget – only 1.39 million iPhones were sold in its first year. It took five years before Apple was selling over 100 million smartphones a year. It wasn’t the device itself that was so remarkable but the suite of innovations that transformed all smartphones in the years that followed.
And how all those millions of smartphones changed us.
In a fascinating article in The Information, Ryan Broderick picks up on the parallels between the introduction of the iPhone and the arrival of ChatGPT.
The world has changed so much since the dawn of the iPhone that we no longer remember how unfulfilling the first generation of mobile devices actually was. If you’re old enough to have used the first Android and iOS smartphones, there’s no doubt you remember the very specific feeling of picking them up, thinking they were pretty neat and then very quickly running out of things to do with them. It wasn’t until years later when data speeds, websites and social platforms had all optimized for mobile that we started to figure out what those devices were really good for. AI will be no different.
In fact, many of the criticisms currently leveled against AI were also leveled against smartphones 15 years ago. Schools banned phones in classrooms. We thought that smartphone use might alter our brain chemistry. Now there are fears that students will use ChatGPT to cheat and panic that AI will kill the human urge to create art. Meanwhile, the things we probably should have been worried about never came up at all. Could we have predicted when Twitter launched in 2006 and the first iPhone shipped in 2007 that the two would be used to organize global political movements, topple governments and change culture forever?
Few people saw the fundamental paradigm shift that the iPhone embodied – and how it would transform society. And the same will be true with the release of ChatGPT since it comes at such a difficult time for the tech industry. We’re just coming out of a period of incredible hype where everyone was talking about Web3, Bitcoin, and the Metaverse. Not one of those has significantly impacted our lives, though the collapse of Bitcoin and NFTs have cost some people a small fortune. At conferences last year, we heard speaker after speaker predict how web3, Bitcoin, and NFTs would be transformative. And here we are today, with those developments in shambles and Meta losing billions on a metaverse rollout we have yet to see.
Perhaps we’re simply not in the mood to hear about another paradigm shift. But ChatGPT and its successors will have a revolutionary impact in ways we cannot fathom. It’s like staring at the ocean from the shoreline and trying to understand the depth of the Mariana Trench (almost seven miles or 11 kilometers, in case you’re curious) – but no one even told you it was out there.
ChatGPT is Not the Most Advanced AI Available
Paradoxically, those well-versed in AI developments don’t see ChatGPT as a significant advance in AI. As people in the artificial intelligence community have said, companies and research labs have built data-driven AI platforms for years. Even GPT-3, the foundation on which ChatGPT functions, benefits from the 20-year development of large language models. Transformer architecture was introduced back in 2017 by Google researchers and is now widely used for natural language processing in chatbots and speech recognition systems.
As Yann LeCun, Meta’s chief AI scientist noted:
In terms of underlying techniques, ChatGPT is not particularly innovative. It’s nothing revolutionary, although that’s the way it’s perceived in the public. It’s just that, you know, it’s well put together, it’s nicely done.
But that “nicely done” is the key to why ChatGPT has broken all records for rapid adoption – a million users in five days and more than 100 million users within the first two months. It’s not an advanced AI product but a well-engineered one. Like the early iPhone, it does multiple tasks reasonably well and a few exceptionally well. The public wasn’t waiting for an advanced artificial intelligence platform. Instead, it wanted AI it could use – and OpenAI’s ChatGPT flung open the doors to that groundbreaking opportunity.
ChatGPT and XR
ChatGPT and related developments will have a massive impact on XR. In working in AR and VR, we always said they would only come into their own in the convergence with AI. One of the most impressive virtual experiences we’ve done over our many years with XR was the encounter with Magic Leap’s AI avatar, Mica. Mica used a combination of AI and spatial recognition to understand your interactions with her, and it was uncanny. You couldn’t help but interact with her as a natural person. But what was so challenging to implement five years ago will become remarkably easy in the near future. Now that ChatGPT is here, we’re already seeing the development of AI-driven avatars by VirtualSpeech and other companies. And we have no doubt it will be incorporated into AR and smart glasses in the coming year.
ChatGPT’s ability to answer your questions, summarize text, and create an outline for you are the tip of the iceberg. They’re the surface features of user-friendly, accessible AI that will profoundly disrupt how we learn, work, and entertain ourselves.
The Impact of ChatGPT and Education
ChatGPT is already significantly impacting educational institutions that have traditionally relied on written work for learning activities and assessments. Student emails are arriving with much more polish and fewer errors. Short assignments are suddenly well-written and logically structured. Major school systems such as Oakland, Seattle, and here in New York City have blocked access to it through school networks – though that’s a laughable solution since students will simply use their phones.
And it’s taken a more serious turn. Researchers are discovering that human reviewers can easily miss abstracts generated by AI. Artists and other creatives are already undertaking legal action as they see their work freely appropriated by related AI platforms such as DALL-E and Midjourney. This isn’t the extent of the disruption, but just the beginning.
Welcome To The Future
Speaking of the future, OpenAI isn’t resting on its laurels. A new version of ChatGPT will arrive this spring. While there’s been much speculation about its features, we’ll just say it will be much more than an incremental advance over the current version. We’re at the threshold of a new era in AI that’s accessible and usable by the public. Compared to the iPhone, it’s profoundly democratic and available to anyone with a basic internet-connected device, be it a phone or laptop. And it’s not just a question of addressing this paradigm shift in the tech community or in the U.S. and Europe. We need a diversity of voices from around the world, as this will impact all of humankind.
Sam Altman, OpenAI CEO, said the system is simply “an early demo of what’s possible.” Think about that: the ChatGPT we are using today is nothing more than an early demo of what the future will bring.
He continues,
Soon you will be able to have helpful assistants that talk to you, answer questions, and give advice. Later you can have something that goes off and does tasks for you. Eventually you can have something that goes off and discovers new knowledge for you.
The iPhone let us walk around with the Internet in our hands, constantly accessible. From Altman’s perspective, ChatGPT may ultimately let us walk around with a second brain in our heads. Of course, his optimism – typical of tech people – raises a host of unanswered questions. To be answered by . . . us?
Who determines and maintains the ethical guardrails, so it doesn’t replicate the worse elements of human nature as other public AI platforms have done? What do we do about the impact on the workforce and creatives in our society? Will it mark the end of programmers, playwrights, professors, journalists, and others? Quite possibly, a more advanced version of ChatGPT could become the rust-belt moment for the educated class in the future.
And what happens a decade or two down the road if the evolution of AI accelerates? If AI ultimately can go off and discover knowledge for you, in the end, is it only finding what it has already created? Does it become a vicious circle of AI talking to itself while humans sit on the sidelines, deluded by the illusion of agency?
We stand at the threshold of a new era, a pivotal moment that will profoundly reshape what we know, what we do, and how we engage with each other. It will be an era filled with abundant opportunity and fraught with the potential for failure.
Welcome to the future.
Emory Craig is a writer, speaker, and consultant specializing in virtual reality (VR) and generative AI. With a rich background in art, new media, and higher education, he is a sought-after speaker at international conferences. Emory shares unique insights on innovation and collaborates with universities, nonprofits, businesses, and international organizations to develop transformative initiatives in XR, GenAI, and digital ethics. Passionate about harnessing the potential of cutting-edge technologies, he explores the ethical ramifications of blending the real with the virtual, sparking meaningful conversations about the future of human experience in an increasingly interconnected world.