“Until the early 2000s, paper maps did not exist — they were simply called maps” wrote Reid Hoffman and Greg Beato in “Informational GPS,” one of 12 essays in The Digitalist Papers, a roadmap of the possible futures that the AI revolution might produce. Before May of 2000, the Global Positioning System (GPS) was deliberately limited for civilian use due to security concerns. But then access to GPS was made available to everyone as a public good
“In the wake of this policy shift, GPS experienced a Cambrian explosion of innovation.” Improved performance and falling prices for consumer receivers quickly turned GPS into an indispensable part of life in the 21st century. A 2019 study on the Economic Benefits of GPS by the National Institute of Standards and Technology (NIST) estimated that by 2017, GPS had generated roughly $1.4 trillion in location-based services across a variety of industries since being made available for civilian and commercial use.
The term GPS has entered the popular vernacular to mean one’s specific location at a specific point in time. Its breakthrough application has been turn-by-turn navigation, enabling us to move through the physical world with precise, constantly updated information. “At literally every turn, these navigation systems increase individual agency by telling us where we are, what else is nearby, what obstacles might impede our progress, and so much more.”
A major reason why GPS has been so widely adopted is that it offers us individual agency along with broad access to navigation information. GPS not only suggests the best routes for us to follow based on current traffic conditions, but it also offers us the ability to modify our route along the way and quickly readjusts to accommodate our wishes and new position.
AI: The Informational GPS Revolution
In their essay, Hoffman and Beato argue that the story of GPS is particularly relevant to AI’s ongoing development. At its core, GPS tells us where we are, what's nearby, and how to get to where we want to go. AI does something similar for our informational journeys. Need to summarize a dense legal document?, understand a complex scientific concept?, plan a new project?, — AI tools can guide us, breaking down barriers of expertise and enabling us to navigate challenges with greater confidence.
Large Language Models (LLMs) and their associated applications function similarly to GPS. “They increase our capacity to navigate the complex and ever-expanding informational environments that define life in the 21st century. In doing so, they enhance the individual agency of billions of people worldwide, by providing the kind of situational fluency that enables higher engagement and more informed decision-making.”
But, there are key differences. GPS deals primarily with objective data, i.e., geographic coordinates and precise timestamps. LLMs, on the other hand, process and generate context-dependent subjective information that is rooted in the nuances and complexities of human language. Unlike GPS, LLMs are not based on a single source of objective data that’s the same for every user. Instead, every LLM is based on a unique “information planet,” that is, on its underlying AI model, the number of parameters in its neural network, and the data used in its training.
The Importance of Broad Access
There are a number of reasons why the story of GPS is relevant to AI’s ongoing development. As discussed earlier, GPS offers broad access to navigation information to billions around the world, as well as the freedom to choose and modify the route they want to follow. Similarly, Hoffman and Beato emphasize the importance of broad access and individual agency in realizing AI’s potential benefits. “They compare AI to GPS technology and propose strategies to develop equitable and inclusive AI systems that build societal trust and deliver benefits to billions of people.”
In addition, as is the case with GPS, an AI that is broadly distributed and offers individual agency “stands as a clear example of the positive outcomes that can result when the government embraces a pro-technology, pro-innovation perspective and views private-sector entrepreneurship as a strategic asset for achieving public good.”
“Fully applying AI to grand challenges like sustainable energy abundance, drug discovery, and more equitable access to healthcare and education will require more than just technological breakthroughs. Broad societal understanding, trust, and a sense of shared purpose matter, too. One way to pursue such ends is through applications that enable individual users to access and experiment with AI directly. When people experience the benefits of new technologies in hands-on, self-determined ways, a sense of equity accrues.”
Challenges and Risks of AI
New, large scale LLMs are being released just about every month. These AI models demonstrate capabilities unimagined a decade ago, but they sometimes generate responses which contain false or misleading information presented as facts, as well as being prone to bias and to their use for negative purposes. While including guardrails to better align them with human values, LLMs can often still be coaxed into producing toxic, discriminatory, or harmful outputs.
In their essay, Hoffman and Beato warn about an even more insidious risk, — AI as a kind of 21st century Big Brother. As we continue to make progress on challenges like model accuracy and bias, concerns over individual agency will likely intensify. “Improvements in reliability and fairness will presumably make models more trustworthy and more authoritative — so much that we begin to cede more and more of our decision-making to AI systems of various kinds. Or others may do so, ostensibly on our behalf but often without our consent or even our knowledge.”
Building a Trustworthy and Inclusive AI Ecosystem
“The 20th century brought us innovations like email, hyperlinks, search, and emojis in response to new informational demands,” wrote the authors in conclusion. “The 21st century has given us AI.”
Over time, most everyone will benefit from AI, just as they do from GPS and smartphones. Making LLMs and other AI applications more inclusive will be an ongoing process. We must develop AI models that are broadly accessible and learn how people use them, where issues arise, and the public beliefs regarding how models should function. “By helping people see what’s in it for them, they’ll also be more likely to appreciate what’s in it for all of us.”
“To ensure that we develop AI in alignment with the democratic ideals of self-determination and participatory governance, we must pursue design paradigms that prioritize individual agency and give people hands-on access to tools that they can use in practical, open-ended ways. Conceptualizing LLMs as a form of informational GPS provides a model for doing that.”
“For the last 20-plus years, however, the prevailing story of GPS is not that its limitations, flaws, and vulnerabilities have led to significant disruptions in security, navigation, and other essential services. Instead, it’s that GPS has globally enabled a wide range of massively beneficial services, every minute of every day, week after week, year after year. With generative AI models that function as a new form of informational GPS, we’re now on the same path. It’s a journey that will give billions of people new powers to navigate the 21st century.”
Comments