AI Can Assist Intelligence, But It Cannot Replace Humans

By: Kharis Kurnia, S.E., M.Si.
Lecturer in International Relations at Bangka Belitung University, International Security Activist

JurnalPost.com - Every time I discuss security and intelligence issues, there is almost always one question: will the job of an intelligence analyst eventually be taken over by AI? This question is natural amidst such rapid technological development. However, in my opinion, the more important question is not whether AI will replace humans, but how humans can use AI without losing their critical thinking skills.

This article is a development of my research titled “Artificial Intelligence, Imagery Intelligence, and Early Warning Systems: A Strategic Intelligence Model for Addressing Contemporary Security Threats.” That research serves as a foundation for examining how artificial intelligence, Imagery Intelligence (IMINT), and Early Warning Systems can be placed within a strategic intelligence framework to address contemporary security threats.

Machines That See Faster, But Do Not Necessarily Understand

Imagine the job of an intelligence analyst who must sift through thousands of satellite images, drone footage, or other visual data every day to look for changes that might indicate a threat. Such work demands extraordinary precision and significant time. As the volume of data continues to grow, we must honestly admit that human capacity to handle it, no matter how skilled an analyst is, still has its limits.

At this point, AI shows its most obvious advantage: speed. In the context of Imagery Intelligence or IMINT, AI can process large amounts of visual data, recognize patterns, classify objects, and flag changes that might escape manual observation. For an analyst, this is like having an assistant that never tires of filtering through piles of data before the relevant parts are handed over for further inspection.

However, there is one thing that I believe we must constantly remind ourselves of, both as academics and as security issue activists: the ability to see is not the same as the ability to understand. An AI system might be able to detect a new object, a change in building structure, or an increase in vehicle activity at a location. But this system is not necessarily capable of answering a much more fundamental question: what is the actual meaning of those changes?

Because not all changes are threats. The problem becomes even more complicated when we talk about speed versus truth. The speed of AI is indeed impressive, but it will turn into a serious problem if the results of its analysis are treated as absolute truth without a verification process. In security matters, such errors are not just simple technical mistakes. These errors can influence threat perceptions, shift policy priorities, and even trigger unnecessary responses.

There is also a danger that I think often escapes attention: data manipulation. No matter how great an AI model is, it is only as good as the data it is based on. If that data has been manipulated, or if the system itself is attacked and disrupted, then the conclusions produced will also be flawed. In other words, the challenge of intelligence in the AI era is not just about making machines smarter, but also about ensuring that the information read by the machine is truly trustworthy.

From here, we can draw an important distinction between data, information, and intelligence. Data can be vast, and information can be processed very quickly. But intelligence demands something much deeper: interpretation and assessment.

An analyst is not enough just knowing that a change has occurred, but must also keep asking why that change happened, whether there is a pattern behind it, how it relates to other information, and what the possible consequences are. That is why I always say we must understand that the results of AI work should be treated as an early signal, not as a final decision.

The Future of Intelligence: Humans Remain in Control

If so, what is the actual position of humans in the midst of all this sophistication? In my opinion, the answer lies in the human-in-the-loop principle, which is ensuring that humans remain in the analysis process chain and have the authority to examine, question, or even reject the results provided by the system.

This principle is by no means a sign that AI is still weak. On the contrary, human-in-the-loop is a form of safeguard so that the power of technology does not turn into a new source of vulnerability.

Within this framework of thinking, AI is best positioned as a force multiplier for human capacity. AI takes over repetitive tasks, processes data on a large scale, finds patterns, and helps determine which parts need further inspection.

With that, human analysts have more time to work on things that are truly of strategic value, such as connecting various sources of information, understanding context, testing assumptions, assessing risks, and considering political and security consequences.

Machines and humans, in my view, are not replacing each other, but complementing each other. Machines excel in speed and scale, while humans excel in context, judgment, and strategic consideration.

For Indonesia, this development has significant meaning. We live in a complex strategic environment, with rapidly moving geopolitical dynamics, while the volume of information that must be monitored continues to grow along with technological developments. In such a situation, the ability to detect changes early is clearly increasingly important.

But I want to emphasize one thing: investment in technology alone will not be enough. The development of AI capacity in the intelligence field needs to go hand-in-hand with improving the quality of human analysts, data governance, system security, verification mechanisms, and clear decision-making procedures.

We must not get trapped in a situation where the system is capable of generating thousands of alerts in a short time, but we do not have enough human capacity to determine which alerts are truly important. A system that is too sensitive will produce too many alarms, and in the end, it will actually make its users exhausted until they start ignoring the warnings.

Conversely, a system that is too conservative risks missing threats that are actually crucial. At this point, human quality becomes the deciding factor again. Technology can speed up the process, but humans must determine how those results are understood and used.

Ultimately, the debate about whether the future of intelligence will be dominated by humans or machines is not the most productive question. What is more important is building the right relationship between the two.

AI can help intelligence see more, faster, and on a wider scale. But humans must still understand what is being seen, test its truth, and determine the consequences.

The more sophisticated AI becomes, the more important our ability to think critically becomes. AI can help intelligence see threats faster, but only humans are capable of giving strategic meaning to what they see.

The future of intelligence is not humans versus AI, but humans who are capable of using AI without surrendering entirely to the machine.