During his keynote talk at the ASU+GSV Summit in San Diego, Gates enthusiastically endorsed the potential of AI chatbots such as OpenAI’s ChatGPT and Google’s Bard. These chatbots have made remarkable strides in recent months, matching or even surpassing human-level intelligence on certain standardized tests. This progress has generated both excitement about the technology’s possibilities and concerns regarding potential negative consequences.
Count Gates among the impressed individuals who believe that today’s chatbots exhibit “incredible fluency at being able to read and write.” He predicts that this advancement will soon enable them to guide students in improving their reading and writing skills in ways that were previously inconceivable. Gates expects the initial impact to be felt most strongly in reading, as chatbots serve as research assistants and provide valuable feedback on writing.
Gates acknowledged the historical challenges computers faced in teaching writing skills. When grading essays, human teachers evaluate elements like narrative structure and clarity of prose—an intellectually demanding task that proves formidable for developers to replicate in code. However, proponents argue that the ability of AI chatbots to recognize and reproduce human-like language alters this landscape significantly.
Kevin Roose, a tech columnist at The New York Times, shared his positive experience using ChatGPT to enhance his writing skills. By leveraging the chatbot’s remarkable ability to swiftly search through online style guides, Roose was able to refine his own writing. Additionally, academics have praised chatbots for their capacity to summarize and provide feedback on text, and some have even witnessed them writing complete essays.
Nevertheless, these academics caution that the technology is still in its nascent stages and may inadvertently introduce significant errors or misinformation. Before AI chatbots can truly become effective tutors, they must improve their ability to comprehend and reproduce human language, thereby better motivating students.
Gates envisages a future where AI chatbots will serve as valuable teacher’s aides, providing feedback on writing tasks in the next 18 months. Subsequently, they will amplify their impact on mathematical education. The revelation that chatbots might excel in reading and writing before mathematics may come as a surprise, given that algebra and calculus often serve as foundational elements for AI development.
However, experts point out that chatbots, which rely on extensive datasets for training, often struggle with mathematical calculations. While a chatbot can provide an answer if a solved math equation already exists within its training data, it faces challenges when tasked with computing solutions independently. Gates frequently questions Microsoft’s AI developers about the limitations preventing chatbots from performing relatively simple calculations or even multiplying numbers. The response consistently highlights the need for improved reasoning capabilities within AI to handle the complexities of mathematical calculations.
Gates remains confident that, although progress may take time, the technology will undoubtedly improve within the next two years. The advent of AI-driven tutoring could potentially make private tutoring accessible to a broader range of students who may otherwise struggle to afford it. However, he clarifies that although chatbot platforms like ChatGPT and Bing currently offer limited free versions, ChatGPT introduced a subscription plan called ChatGPT Plus in February, priced at $20 per month.
Nonetheless, Gates emphasizes that AI tutoring will be comparatively more affordable and accessible than traditional one-on-one tutoring with a human instructor. The implementation of AI tutoring stands to level the playing field, as access to a tutor has traditionally been prohibitively expensive for most students. Moreover, AI tutors possess the unique ability to adapt to individual needs.