Have you ever seen U.S. Customs and Border Protection agents working with dogs to scan an airport for drug couriers?
The agents have been trained to look for certain behaviors and characteristics. For people who buy tickets last minute, in cash, to countries known for drug production. For heavy, bulging suitcases. For travelers flying alone, who stare at agents’ badges while constantly licking their lips. The dogs, meanwhile, undergo a months-long program to detect thousands of different scents.
Agents are good at some tasks, dogs at others. Together, they catch more drug couriers than either could on their own.
The best AI works the same way — as a different type of intelligence that complements and bolsters human capabilities, empowering us to complete a variety of tasks better and faster.
As Babel Street debuts Insights GPT Beta, I wanted to take a moment to discuss artificial intelligence: what it is, how it works, and how Insights GPT Beta and similar programs mark the first steps toward true human-machine collaboration.
What is Insights GPT Beta?
Insights GPT Beta is a generative AI tool that empowers customers to ask questions of their data using everyday language. The “GPT” stands for “generative pre-trained transformer.” This is a type of computing model that studies and learns from extraordinarily large language samples to understand human queries posed in plain language, and to provide responses written in plain language.
Conversational search tools now in general use base their responses on a variety of sources — Wikipedia, books, news articles, general websites, and more. When you ask a conversational search tool, “What does U.S. Customs and Border Protection use dogs for?,” it responds with information that it gleans from these sources. For example, it may answer, “U.S. Customs and Border Protection (CBP) uses dogs, specifically trained as detector dogs, for various purposes related to border security and law enforcement …" (See Figure 1,)
But Insights GPT Beta does more than present information from commonly available sources. Rather, working with the Babel Street Insights big data platform, it also scours the Babel Data Library — the world’s largest and most diverse collection of publicly available and commercially available data. Outside the library, it further searches real-world interactions generated on message boards and social media — including blogs, posts, and online comments; and information published on the deep and dark web.
Insights GPT Beta both hastens time to insight from all this data and improves the quality of insight obtained. It provides a new layer of essential analysis for organizations that require rapid, contextual understanding for advanced intelligence analysis, identity resolution, detection of financial crimes, risk mitigation, and other tasks. Advanced summarization capabilities deliver concise insights and analyses of large volumes of information. New filtering capabilities help you select information from your most trusted data sources.
Most significant, Insights GPT Beta empowers you to use chat-like capabilities to target your research: asking specific questions and receiving specific answers — more easily distilling and refining queries to obtain the information that matters most to you.
Future Capabilities of AI-powered Language Tools
As noted earlier, conversational search is just the first step in human/AI collaboration. What data insight will AI be able to deliver in the future?
Scenario simulation is one likelihood. This human-AI collaboration capability will one day help you answer “what if” questions surrounding potential future events.
Think about U.S. Customs and Border Protection. Officials at that agency could use scenario simulation to answer questions such as, “What if we provided 1,000 more agents with detector dogs? How many more drug couriers could we apprehend? What volume of drugs could we stop from entering the United States? At what airports should we place these dogs for maximum impact?”
In the financial field, analysts will be able to use scenario simulations for tasks such as determining ultimate beneficial ownership of businesses. In some cases, business owners are hidden by complex corporate structures: An LLC in one country is owned by a C-Corp in another country, which is in turn owned by another corporation in a third nation. These types of complicated structures are sometimes used for money laundering and other criminal purposes. By enabling users to uncover potential relationships between and among entities (“What if the CEO of company A is the brother-in-law of the CFO of company B?”) scenario simulation will help financial institutions determine who owns the companies they do business with. This will empower them to better comply with anti-money laundering statutes.
AI and Professional Superpowers
Capabilities such as conversational search and data summarization will help us to excel in our jobs. They will give us professional superpowers.
As discussed earlier in this blog, the best AI does not replicate human expertise. Rather, it’s engineered to complement human capabilities, to excel where we struggle. Already, AI is helping software engineers to complete code faster. In regulated industries, AI is helping us determine pathways for assessing risk.
In finance, we know that criminals worldwide launder between $800 billion and $2 trillion each year, representing anywhere from 2% to 5% of the world economy. Instances of fraud are both increasing in number and growing more expensive to correct. (Each $1 of fraud loss now costs FIs $4 to resolve, according to the “LexisNexis® True Cost of Fraud™ Study: Financial Services & Lending.” ) We catch too little of either. But AI’s ability to trawl through millions or billions of data points, finding and recognizing patterns, gives it an edge for spotting possible instances of fraud and money laundering. Humans can then devote their time to determining what’s truly fraudulent and finding ways to stop these activities.
In these areas and more, AI is becoming our partner. And people seem to enjoy the relationship. Sixty-four percent of the employees surveyed in the “MIT Sloan/BCG 2022 Artificial Intelligence and Business Strategy Global Executive Study and Research Project” say they benefitted from human-AI collaboration, and that they were at least three times more likely to be satisfied with their jobs after working with AI. Sixty percent of these respondents said they already felt that AI was more of a coworker than a tool or a threat.
Soon, AI will be part of every application, and embedded in most professional tasks. Fewer jobs will be completed by computers alone, or by humans alone. Most will be a human-AI collaboration. Interacting with an other-than-human intelligence will help each of us bring something unique to our work. Just ask U.S. Customs and Border Protection agents how much they value their dogs.
1. United Nations Office on Drugs and Crime, “Money Laundering,” accessed May 2023, https://www.unodc.org/unodc/en/money-laundering/overview.html
2. Lexis Nexis, “LexisNexis® True Cost of Fraud™ Study: Financial Services & Lending,” January 2022, https://risk.lexisnexis.com/about-us/press-room/press-release/20220106-annual-true-cost-of-fraud-study
3. MIT Sloan Management Review, “Achieving Individual — and Organizational — Value with AI: Findings from the 2022 Artificial Intelligence and Business Strategy Global Executive Study and Research Project,” October 2022, https://sloanreview.mit.edu/projects/achieving-individual-and-organizational-value-with-ai/