The last few years—even the last few months—have seen artificial intelligence (AI) breakthroughs come at a dizzying pace. AI that can generate paragraphs of text as well as a human, create realistic imagery and video from text, or perform hundreds of different tasks has captured the public’s attention. People see AI’s high level of performance, creative potential and, in some cases, the ability for anyone to use them with little to no technical expertise. This wave of AI is attributable to what are known as foundation models.

What are foundation models?

As the name suggests, foundation models can be the foundation for many kinds of AI systems. Using machine learning techniques, these models apply information learned about one situation to another situation. While the amount of data required is considerably more than the average person needs to transfer understanding from one task to another, the result is relatively similar. For example, once you spend enough time learning how to cook, without too much effort you can figure out how to cook almost any dish, and even invent new ones.

This wave of AI looks to replace the task-specific models that have dominated the landscape. And the potential benefits of foundation models to the economy and society are vast. For example, identifying candidate molecules for novel drugs or identifying suitable materials for new battery technologies requires sophisticated knowledge about chemistry and time-intensive screening and evaluation of different molecules. IBM’s MoLFormer-XL, a foundation model trained on data about 1.1 billion molecules, helps scientists rapidly predict the 3D structure of molecules and infer their physical properties, such as their ability to cross the blood-brain barrier. IBM recently announced a partnership with Moderna to use MoLFormer models to help design better mRNA medicines. IBM also partners with NASA to analyze geospatial satellite data—to better inform efforts to fight climate change—using foundation models.

However, there are also concerns about their potential to cause harm in new or unforeseen ways. Some risks of using foundation models are like those of other kinds of AI, like risks related to bias. But they can also pose new risks and amplify existing risks, such as hallucination, the capability of generation of false yet plausible-seeming content. These concerns are prompting the public and policymakers to question whether existing regulatory frameworks can protect against these potential harms.

What should policymakers do?

Policymakers should take productive steps to address these concerns, recognizing that a risk and context-based approach to AI regulation remains the most effective strategy to minimize the risks of all AI, including those posed by foundation models.

The best way policymakers can meaningfully address concerns related to foundation models is to ensure any AI policy framework is risk-based and appropriately focused on the deployers of AI systems. Read the IBM Policy Lab’s A Policymaker’s Guide to Foundation Models—a new white paper from us, IBM’s Chief Privacy & Trust Officer Christina Montgomery, AI Ethics Global Leader Francesca Rossi, and IBM Policy Lab Senior Fellow Joshua New—to understand why IBM is asking policymakers to:

  1. Promote transparency
  2. Leverage flexible approaches
  3. Differentiate between different kinds of business models
  4. Carefully study emerging risks

Given the incredible benefits of foundation models, effectively protecting the economy and society from its potential risks will help to ensure that the technology is a force for good. Policymakers should swiftly act to better understand and mitigate the risks of foundation models while still ensuring the approach to governing AI remains risk-based and technology neutral. 

Read “A Policymaker’s Guide to Foundation Models”

More from Artificial intelligence

The five key benefits of AIOps and automation

4 min read - If you’re an IT professional—from the C-suite to a hands-on practitioner—you know the pressure your IT operations (ITOps) are under. You’re responsible for optimizing spend, operational efficiency and incorporating new and innovative technologies. But are your tools slowing you down? Coined by research firm Gartner, AIOps is artificial intelligence for IT operations. It is the application of artificial intelligence (AI) capabilities (e.g., natural language processing and machine learning models) to automate and streamline operational workflows. In this blog post, we…

4 min read

What is smart transportation?

5 min read - Every day, people encounter multiple obstacles while traveling to their intended destinations. Sitting in traffic, waiting for the bus to arrive 15 minutes later than scheduled, driving around for 30 minutes to find a parking spot—the modern world is full of inconveniences due to underlying inefficiencies in our transportation systems. However, stalled cars and harried people waiting for public transportation aren’t just an individual nuisance. A less-than-optimal transportation infrastructure affects the economy, hastens environmental impact and lowers the overall quality…

5 min read

IBM named a Leader in the latest Forrester Wave™ report for AI Decisioning

2 min read - Forrester Research just released "The Forrester Wave™: AI Decisioning Platforms, Q2 2023: The 13 Providers That Matter Most And How They Stack Up" by Mike Gualtieri with Aaron Katz, Catherine Marcin, and Jen Barton, and IBM is proud to be recognized as a Leader.  This report outlines the combination of traditional decision automation tools with machine learning models and other technologies. As Forrester notes in the report, many organizations are eager to harness the power of AI but also must…

2 min read

Capgemini and IBM Ecosystem strengthen partnership for Drone-as-a-Service

5 min read - Today, utilities and many other industries use drones extensively to conduct surveys, map assets and monitor business operations. They use drones for tasks as simple as aerial photography or as complex as sophisticated data collection and processing. The global commercial drone market is projected to grow from USD 8.15 billion in 2022 to USD 47.38 billion by 2029, at a CAGR of 28.58% in the forecast period. Different drone requirements require different investments such as commercial drone pilots with specialized…

5 min read