Last week, I participated in the President’s National Artificial Intelligence Advisory Committee (the “NAIAC”) as chair of the Trustworthy AI Working Group. ChatGPT and generative AI was a hot topic of discussion. I wanted to set out some historical context for ChatGPT as well as some of BSA’s thoughts on developments in responsible AI.
First, a brief historical context and some milestones along the way:
1940s: Neural networks
In 1943, the mathematical theory behind neural networks (networks of artificial neurons that can perform logical functions) was published by Warren McCulloch and Walter Pitts.
1950s: “AI” research takes off
John McCarthy coins the term “artificial intelligence” in 1955 and, in 1956, McCarthy organizes the Dartmouth University conference: experts discuss neural networks, computer vision, and natural language processing. The Dartmouth Conference marks the beginning of AI as a recognized field of study.
1960s: Natural language processing; the first chatbot
ELIZA, the world’s first chatbot — an early use of natural language processing — is created at MIT. ELIZA used text to communicate in seemingly intelligent conversations.
1980s: AI starts being commercialized
The 1980s marks the beginning of non-tech companies using “expert systems” as companies begin to see the potential of AI to solve problems.
There are also significant strides in the use of “backpropagation” to train neural networks to learn to correct mistakes.
Japan launches the Fifth Generation Computer Systems project, accelerating research on programming machines that could translate, talk, and emulate human reasoning.
1990s: Deep Blue becomes world chess champion
The headline-grabbing ‘90s event was Deep Blue’s 1997 victory over world chess champion Gary Kasparov, capturing the attention of the general population. Deep Blue demonstrated AI’s outstanding ability to solve problems with clearly defined rules using a computer’s high-speed capabilities.
2010s: Natural language processing and neural networks
In 2011, IBM’s Watson demonstrated the capabilities of natural language processing combined with fast processing speeds to become Jeopardy champion. Apple released “Siri,” the personal assistant’s speech recognition and natural language processing, which marked the first widespread use of AI by the general public.
In 2016, DeepMind’s AlphaGo beat the world Go champion. Because of the vast number of allowable board positions (more than the total number of atoms in the universe), AlphaGo could not use brute force; instead, it used neural networks to learn as it played.
Today, AI is transforming every sector of the economy. Manufacturers use AI to design safe and sustainable products; health researchers use AI to drive medical breakthroughs, including to protect against future COVID-19 variants; and companies use AI to make their products more accessible for people with disabilities.
The announcement of ChatGPT has created a surge of interest in AI from both the general public and policy makers. ChatGPT is a large language model, released by Open AI in November, that uses deep learning to generate text that is similar to human language. It has the potential to be used by companies and by individuals on a scale that is unprecedented.
BSA: Leading the Discussion on AI Policy
As the global voice of enterprise software, BSA has been at the forefront of thoughtful policy proposals that promote the responsible development of AI. Key elements include:
- Requiring Impact Assessments for High-Risk AI
- Developing Risk Management Frameworks
- Ensuring Strong Privacy Rules and Trusted Data Transfers
In 2021, we released a framework for organizations to implement to reduce the risk of bias in AI: “Confronting Bias: BSA’s Framework to Build Trust in AI.”
You can find our latest thinking in BSA’s 2022 Artificial Intelligence Policy Agenda, which provides five key pillars for enabling and promoting responsible AI.
Please reach out to us with questions or thoughts. The importance of responsible AI will only continue to increase for our economy and our society as we move forward.
