Building Better AI: The Need for R&D Investment in a Crowded Market
by: Mitchell Sutika-Sipus, Chief Solutions Architect
It is easy to build an A.I. company these days. Anyone can do it. The public explosion in Large Language Model (LLMs) in the last few years has made it possible for anyone to cobble together software, find customers, and market solutions. This has spurred the growth of many new markets too. For example, using AI to write government proposals is a new, yet suddenly crowded market. Worse, a proposal workflow slapped on top of Chat-GPT will not win the contract. Instead it will absorb your time and money, then leave you worse off.
In this crowded market, there are only a few serious companies like AutogenAI, who supply AI in a highly specialized manner to create the change you want to see in the world. We invest in internal research teams to build complex solutions that solve complex problems for our customers. Unfortunately for our customers, most of our competitors generate low quality solutions. The magic of large language models sits within the illusion of human-like understanding and dialogue, and this illusion works well because LLM’s are designed to generalize information. If you want concise, high fidelity and nuanced writing – most of these companies disappoint you.
An economy built on low-quality, non-differentiated technologies is not set for growth. It may grow immensely in the early days, but the decline is surely steep when everyone who bought these AI tools is suddenly less successful than before. Great companies in the United States are founded not just on ambition or reach but are driven by effective Research & Development.
Unfortunately, at the same time as the LLM explosion, we have noticed that most startups are crippled in their ability to differentiate themselves. IRS Tax Code 174 does not allow small businesses to deduct the expense of software development for Research & Development. It only permits annual deductions for granular upgrades with commercial solutions. This means all the new startups who really want to do something new – with the best minds and the best teams – are penalized. Historically a new company trying to invent something new – with no sales and operating purely on venture capital – could claim zero income. Although today, under Section 174, that company must consider the investment as revenue, and pay taxes on all their research in that first year of operation. They are undermined in their ability to differentiate themselves.
This hurts everyone. AutogenAI has thus advocated to The White House Office of Science and Technology Policy for a reconsideration on the role annual expensing for Research & Development for A.I. technologies. To ensure the US remains a leader in A.I. technologies, it is vital that the US economy is supported by a diversity of quality solutions.
