The future of AI programming is undergoing a significant transformation, moving away from complex coding and embracing simplicity. According to Adam Wertheimer, a visiting scientist at the Broad Institute and former Google executive, the days of AI programs consisting of millions of lines of code are coming to an end. Wertheimer believes that with advancements in computing power and access to vast amounts of data, an efficient AI program could be distilled down to just 100 lines of code, or even as few as 10 lines.
Wertheimer’s perspective reflects a broader shift in how AI programs are designed, which is exemplified by the rise of generative AI applications like ChatGPT. Unlike traditional AI systems that relied on intricate coding, generative AI leverages machine learning techniques, particularly neural networks, to learn and perform tasks by analyzing extensive datasets. This change in approach has opened up new avenues for innovation and contributed to the emergence of a potential trillion-dollar industry.
While Massachusetts, particularly the Boston area, has long been associated with AI programming, it has not taken the lead in generative AI development. Recent data from market tracker CB Insights reveals that out of 204 venture-capital-backed startups using generative AI technology, only five are located in Massachusetts. In contrast, California boasts 112 such startups, followed by New York with 37 and Texas with eight. Additionally, none of the 13 startups that have reached unicorn status, with valuations of $1 billion or more, are based in Massachusetts.
The roots of AI programming in Massachusetts can be traced back to the 1950s when researchers at MIT, including Marvin Minsky and John McCarthy, coined the term artificial intelligence. MIT went on to develop the LISP programming language specifically tailored for coding AI systems, with a significant portion of ITA’s software written in LISP. However, the emergence of machine learning and its potential for improving AI remained underappreciated until around 10 to 15 years ago when researchers in New York and Canada made substantial progress in this field.
ITA, a successful travel software startup founded by Wertheimer and his colleagues, utilized a sophisticated junction tree algorithm powered by around 2 million lines of code to search and compare flights. Despite its impressive capabilities, Wertheimer believes that a similar program could now be achieved with just a fraction of that code, thanks to the advancements in machine learning and computing power.
Wertheimer’s journey at Google, where he joined after ITA was acquired, further solidified his belief in the transformative power of machine learning. Over his eight-year tenure, he witnessed Google’s exponential growth from 20,000 to 200,000 employees. While Silicon Valley’s tech mentality demonstrated a greater tolerance for risk-taking, Wertheimer remains rooted in the Boston area, driven by the strong local ecosystems in life sciences and the growing interest in applying AI in various industries.
Boston-based startups like Portrait Analytics, which utilizes chatbot technology to assist financial analysts, are taking advantage of the region’s expertise in healthcare, financial services, and climate tech. However, with established AI companies in Massachusetts like ITA absorbed by Google, iRobot’s acquisition by Amazon, and Nuance Communications being bought by Microsoft, the next wave of AI innovation may be driven more by the West Coast.
Wertheimer’s vision for the future lies in harnessing machine learning to develop new drugs. While the field holds substantial promise, competition among startups and established pharmaceutical companies is already fierce. His new venture aims to leverage the thriving life sciences community in the Boston area and deliver meaningful advancements in drug discovery.
As the AI landscape continues to evolve, it is clear that the industry is undergoing a paradigm shift. The focus is shifting from complex coding to simplified yet powerful programs driven by machine learning and extensive datasets. While Silicon Valley currently dominates the generative AI sector, the vibrancy of Boston’s AI community and its specialized expertise in various sectors hint at the region’s potential to make substantial contributions to the field. As the trajectory of AI programming continues to unfold, it is essential to embrace these advancements while exploring new possibilities and maintaining a commitment to scientific excellence.
