Ark Invest: #344: Tesla’s New 500-Mile-Range Semi Trucks Defy Skeptics, & More
Last week, OpenAI released a conversational AI model called ChatGPT. We asked ChatGPT to create a blog post about itself. In a few seconds, it generated this paragraph:
ChatGPT uses a cutting-edge approach called transfer learning to build upon the knowledge and capabilities of previous language models. This allows it to quickly adapt to new tasks and environments, making it an ideal choice for a wide range of applications. One of the key features of ChatGPT is its ability to understand and respond to natural language input. This means that developers can use it to create AI assistants that can hold a conversation with users in a way that feels more natural and engaging. Another key advantage of ChatGPT is its ability to generate text that is both coherent and diverse. This makes it an ideal choice for applications that require the generation of long-form text, such as chatbots, virtual assistants, and other conversational AI applications. Overall, ChatGPT is a powerful new tool that has the potential to revolutionize the way we interact with conversational AI. Its ability to understand and respond to natural language input, as well as its ability to generate coherent and diverse text, make it a valuable asset for developers looking to create more engaging and realistic conversational AI applications.
In addition to prose, the ChatGPT can write poems, movie scripts, and academic essays. It can explain quantum physics, tutor math, and even write software code. If a user asks it to write a program to process payments on Stripe, for example, ChatGPT will generate a code snippet that communicates with the Stripe Payment API. That said, at this early stage, ChatGPT is prone to “hallucinations” that will require students, programmers, and other authors to check “their work” before considering it complete.
Readers can try the open-access model at:
In our view, AI will transform education and knowledge-work. With ChatGPT, students now can generate essays or code in seconds, and script writers’ movies or plays in minutes. With more iterations and breakthroughs, the possibilities seem open-ended.
According to ARK’s recent research, AI should more than double the productivity of knowledge workers by 2030, potentially boosting real GDP by $56 trillion, but based on recent breakthroughs, that estimate could be too conservative. GitHub Co-Pilot already can double the productivity of software engineers performing coding tasks, and now ChatGPT will turbocharge the progress.