{"id":109,"date":"2025-06-19T22:47:23","date_gmt":"2025-06-20T05:47:23","guid":{"rendered":"https:\/\/blogs.ubc.ca\/scaffoldingchatbots\/?page_id=109"},"modified":"2025-06-21T21:08:05","modified_gmt":"2025-06-22T04:08:05","slug":"technological-trajectory","status":"publish","type":"page","link":"https:\/\/blogs.ubc.ca\/scaffoldingchatbots\/technological-trajectory\/","title":{"rendered":"Technological Trajectory"},"content":{"rendered":"\n

As we just discussed, the AI and intelligent tutoring system market is on a precipitous rise. For our purposes, they are indicators of a significant market opportunity. Chatbots that are able to separate themselves from the pack and deliver a defensible value proposition through enhanced education features have a bright future.<\/p>\n\n\n\n

But the technology itself is evolving at an explosive pace, with market leaders of the core Large Language Models persistently leapfrogging one another in capability, features, scale, and cost. In this section, we will discuss the implications of this rapid technological evolution on the educational chatbot market.<\/p>\n\n\n\n

Moore’s Law No More<\/h2>\n\n\n\n

In 1975, Gordon Moore, the co-founder of Intel, made the observation that the number of transistors in an integrated circuit was doubling roughly every two years (Moore’s Law, 2025<\/a>). This observation of semi-conductor advancement has proven quite accurate, projecting the rise of computer chip capability from room-sized computers to the incredibly powerful computers that fit in our pockets. This observation is now referred to as Moore’s Law.<\/p>\n\n\n\n

\"\"<\/a>
Transistor counts in microchips from 1970 to 2020, demonstrating Moore’s law trend of doubling approximately every two years. From “Transistor Count over time” by M. Roser & H. Ritchie, 2020, Our World in Data (https:\/\/ourworldindata.org<\/a>). Licensed under CC BY 4.0. https:\/\/creativecommons.org\/licenses\/by\/4.0\/<\/a><\/figcaption><\/figure>\n\n\n\n

Modern chatbots are following a very similar trend in the scaling of their capabilities. After decades of comparatively primitive functionality and periodic advancement, beginning in 2010, AI models doubled their capability – not every 2 years – but every 4 to 6 months!<\/p>\n\n\n\n

No one knows whether this technological scaling will hit a ceiling and require a pivot, or continue in perpetuity. <\/p>\n\n\n\n

Reframing Chatbots<\/h2>\n\n\n\n

Under the assumption that this AI scaling and general technology eruption continues, we propose several areas of expansion that long-term investment in the educational chatbot market should take into account.<\/p>\n\n\n\n

Modality<\/h3>\n\n\n\n
\"\"
Eliza, a 1966 Chatbot https:\/\/en.wikipedia.org\/wiki\/ELIZA<\/a><\/figcaption><\/figure>\n\n\n\n

Up until this point, our discussion around chatbots has been centred around the ubiquitous text-based chat interface. That is the dominant modality that chatbot interaction has occurred through since the early chatbots of the 1960s. But this limitation is rapidly being overcome as chatbots expand into other modalities and capabilities.<\/p>\n\n\n\n

Voice<\/h4>\n\n\n\n

Products such as Siri and Alexa brought chatbot interaction outside of text and into the realm of vocal conversation. They were limited though, erring more on the side of voice-command capability than true chatbot conversation. OpenAI’s GPT-4o model brings voice conversation with a chatbot to striking reality.<\/p>\n\n\n\n

\n