Is human intelligence, or HUMINT, losing its relevance as countries’ spy agencies increasingly become technology- or Artificial Intelligence (AI)-driven?
At first glance, the answer seems to be “Yes”, given the way America’s Central Intelligence Agency (CIA) is projecting its ongoing transformation. But, on closer scrutiny, the answer could be “not really”, as the CIA is not replacing field agents but making them “Smart” by endowing them with high-tech tools.
They may now spend less time in alleys and more time as a “data shepherd” — tasking drones, validating AI-generated leads, and operating in the digital-physical seam. But the uniquely human skills — elicitation, empathy, moral judgment — will continue to dominate.
Of course, one sees that more headlines in recent weeks have focused on the CIA undergoing a sweeping structural and operational overhaul under Director John Ratcliffe to prioritize AI, quantum computing, and offensive cyber operations.
Last Tuesday (June 30), addressing a tech conference in Washington hosted by Amazon.com Inc.’s Web Services unit (the AWS DC Summit), Ratcliffe revealed how recent CIA-supported operations in Venezuela and the Middle East, including the rescue of a downed F-15E Strike Eagle pilot in Iran, were examples of the outsized impact of technology on the agency’s intelligence operations.
“It was a search that rested on our innovation, creativity, and our technological know-how, and ultimately it was a technology-enabled search that only the CIA could successfully and did successfully pull off,” the CIA director said of the rescue effort, which he described as “the equivalent of trying to find a needle in a haystack.”
He then added, “Increasingly, all of our future successes are going to depend on technology….. We have to continue to push the boundaries of what’s possible, because the nation that best harnesses the power of technology will determine the global future.”
He promised that the CIA is “going to do everything we can to deliver” the best AI tools to its officers, calling artificial intelligence “a domain in which the CIA must excel, because every algorithmic decision has implications for U.S. strategic advantage and for the national security of all of our people.”
According to Ratcliffe, the CIA has made about 400 technology contracting acquisitions in the last six months, to complete most deals on the same half-year timeline. That, apparently, “would reduce what previously had been an up to 24-month process followed by a nine-month security review”.
Though he did not reveal the exact names of these contractors, he did indicate that talks are underway with Elon Musk’s SpaceX, Amazon, Alphabet Inc.’s Google, and Dell Technologies Inc.
It may be noted that earlier this year, the CIA launched its new commercial technology acquisition framework, led by Chief Procurement Officer Effie Fragogiannis. While still nascent, the framework is said to be delivering on its charge to complete most commercial tech acquisitions in six months or less. Ratcliffe seems happy with its performance of nearly 400 acquisitions, saying that “this is something that previously would have taken several years.”
Accordingly, to provide industry partners with a single point of contact at the CIA, the agency has created an “Office of Corporate Partnerships,” which, Ratcliffe said, is engaging with tech firms.
Incidentally, at its summit, AWS announced a $1 billion incentive program for the intelligence community, which the company said is “designed to eliminate the migration costs that have kept some [intelligence workloads] locked in on-premises systems.”
In a press release, the company said, “The program is simple: migrate qualified workloads to AWS, receive credits. Up to $1 billion is available through October 2030 for all IC agencies on the existing AWS contract.” It added that the CIA, a longtime customer of AWS for cloud technologies, would look to use the program.
In any case, the CIA finds the use of AI indispensable as its OSINT (Open-Source Intelligence) efforts have exploded, driven by commercial satellite imagery, TikTok trends, and dark-web markets. AI tools triage this flood of data streams.
Modern agentic AI pipelines combine Natural Language Processing (NLP), computer vision, and graph machine learning to automate the filtering of global data, flagging anomalies, assessing risks, and generating rapid investigation summaries. AI turns the volume of data into insight in hours, not months, as before.
Secondly, only AI can defend itself against AI in the sense that when adversaries utilize their AI to collect data from or wage an information war against a country, or deepfake its leaders, only its own AI can confront them. It is like a “machine-vs-machine” work – detecting AI-generated personas, hardening communication networks against adversarial multilayered defense, and building “red team” models that try to trick US systems before an adversary does (it involves systematically probing your own systems to identify vulnerabilities, biases, and failure modes before malicious actors can exploit them. By emulating adversarial tactics, organizations stress-test AI models to discover real-world abuse paths).
Thirdly, the American intelligence networks are finding it increasingly difficult (it may be true for others too) to recruit human sources in places like Beijing, Tehran, Moscow, and Pyongyang, thanks to their biometric borders, ubiquitous surveillance, and digital footprints.
But here, the AI, through one space-based support, can still collect critical information from those countries, such as detecting a thermal anomaly at a nuclear site or a surge in GPU imports (Graphics Processing Units, specialized computer chips designed to accelerate computing tasks).
In other words, technology gets you information that humans can’t.
Fourthly, and as a corollary of the above point, if information collected by technology proves false (possibly due to a compromised algorithm), the situation is still preferable to the death of a compromised agent in the field at the hands of the enemy. In that sense, technology reduces physical risk and is cheaper per insight than maintaining a global human network.
However, none of this means that AI will replace human resources at the CIA. All told, the CIA’s 2023 AI strategy explicitly says the goal is “human-machine teaming.”
It makes clear that while AI summarizes intercepts, drafts first-cut cables, and finds non-obvious links between entities, only a well-placed human source can explain what is genuine and what is fake. In other words, while technology can be gamed or manipulated, trust still needs people. After all, AI is excellent at “what” and “where”, but is weak at “why.”
Besides, there are ethical and accountability factors: an AI that hallucinates a source or mislabels a hospital as a weapons depot creates policy disasters. One cannot blame the algorithm for a drone strike on the hospital. That is why a human in the loop is the best legal and moral backstop.
No wonder former CIA Director Bill Burns asserts that human intelligence (HUMINT) remains irreplaceable. While acknowledging the immense value of artificial intelligence and open-source data in finding patterns, he maintains that human agents are still necessary to execute clandestine operations and gather deep, nuanced insights that machines cannot.
Even the present director seems to agree with his predecessor, saying that the CIA would still be driven by human decision-making as it embraces AI, stressing that “only people can decide which is the right way to go.”
Incidentally, under Ratcliffe, the CIA is actively increasing human intelligence operations by aggressively recruiting foreign assets and informants in strategic rival nations, such as China.
“I do not know if this is the best year that the CIA’s ever had, but it’s the best year I can ever remember,” Ratcliffe remarked to Congress during hearings on March 18. Morale in the CIA was high, he insisted. “Human sources, the agency’s bread and butter, were up by 25%, as was foreign-intelligence collection overall. Collection on China had doubled”, he said, adding, “It is a workforce that knows it is doing a great job.”
In sum, then, the CIA is augmenting its officers, not replacing them.
- Author and veteran journalist Prakash Nanda is Chairman of the Editorial Board of the EurAsian Times and has been commenting on politics, foreign policy, and strategic affairs for nearly three decades. He is a former National Fellow of the Indian Council for Historical Research and a recipient of the Seoul Peace Prize Scholarship.
- CONTACT: prakash.nanda (at) hotmail.com
