Traditionally, OSINT has been the application of intelligence methods and tradecrafts to openly available information. Central to the intelligence method is the concept of the intelligence cycle, which describes the usual process to produce intelligence. This model puts a premium on human expertise at every stage of the cycle. To define relevant indicators and scenarios and information requirements one needs to understand the specifics of a particular region or issue.
Why is human analysis critical to OSINT?
Although standard indicators have been developed to forecast traditional risk (political instability, civil unrest, terrorism, or war), each country and situation is different, and it is necessary to take into account those local idiosyncrasies to outline an effective intelligence analysis.
The same goes for collection and assessment. Sources are assessed through a mix of objective metrics (reach, access…) and subjective ones (bias, reliability…) which require the knowledge and experience of a human analyst.
Some technology companies claim that they can produce intelligence analysis automatically using artificial intelligence. However, a case where such solutions actually produced an analytical output a decision-maker can use to make a decision is yet to be seen. The fact of the matter is that human analysts will remain critical to every stage of the work for the foreseeable future.
Human analysis is necessary because only humans:
- can currently compute many different factors in analytical meaningful ways
- have the creativity to tackle new phenomena or question and
- can understand the significance of information in context.
What is the role of technology in OSINT?
OSINT in the 21st century cannot only rely on human insights, expertise, and creativity. The sheer volume of information constantly produced means technology must be included at every stage of the intelligence cycle.
Besides, technological development is changing the intelligence model. It used to be a push system where intelligence is selected, assessed, and analysed by an intelligence professional before being fed to decision-makers. Now it is moving toward a pull system where decision-makers will pull the intelligence they require "on-demand". But the production of such intelligence on demand is not an easy task. One needs to develop highly complex information and intelligence technological infrastructure.
How is technology used by analysts?
Over the past 20 years, the number of technological solutions to help with the various stages of intelligence has exploded. From free to use software like TweetDeck to high-end solutions such as Palantir, technologists have developed hundreds of more or less complex and specialised pieces of software to improve the tradecraft.
The tech stack available to OSINT analysts ranges from the easy-to-use software to technology requiring extensive technical expertise. As a result, OSINT analysts are increasingly split between the domain and technical experts.
While such specialisation is a sure sign that the tradecraft is becoming more mainstream and mature, it is also raising issues about how to integrate those technologies with human expertise.
Indeed, many technical solutions have been developed from the technical experts' perspective rather than from the domain expert one. And many of those solutions have proven highly valuable to the domain experts. However, there's often some confusion about what can be easily done technologically speaking and what is valuable for domain experts. What's more, the relative lack of technical skills among the domain experts has hindered the domain experts-led technological development.
Reconciling human expertise with technological capabilities has increasingly become a priority both for national agencies and private intelligence providers.
This realisation that the human expertise will remain critical to produce good intelligence became particularly obvious in the wake of the failed promises of AI for intelligence. From conversations with intelligence practitioners in the past couple of years, it has become clear that AI is nowhere near what is needed to support intelligence analysis.
As national agencies and private companies have come to realise that their expensive human workforce is here to stay they are looking for technology that empowers their analysts to do their job more efficiently.