In June 2025, the Criminal Investigation Department (CID) of Pune initiated a pilot project by launching AI-based data measurement system. The system collates biometric data of criminals, both accused and convicted individuals. It also collects photographs of arrested persons from 12 different angles, fingerprints, iris scans and other features apart from samples of DNA fingerprints. By storing the data in NIC’s cloud server accessible from across the country, the AI system help police track down criminals just by feeding the modus operandi of the crime. Criminals tend to repeat their crimes in a pattern. Often, they have specific skills and traits that set them apart in the crime scene. The AI will automatically match these patterns to enable police in rounding up the suspects. Even if the criminal alters his look with changing his hairstyle or weight, the AI will successful identify the individual. In the Ganesh Chaturthi festival in Pune, the AI surveillance system flagged 50 instances of abandoned or suspicious bags and issued 250 alerts related to women found in isolated or potentially unsafe locations. AI-powered policing is helping law enforcement agencies in predicting, preventing and investigating crimes.
AI in crime scene analysis
The first step in criminal investigation is analysis of the crime scene. Crime scene analysis helps investigators gather evidence, piece together the chain of events, estimate the potential suspects and motive. An effective crime scene analysis involves gaining an accurate understanding of the victim’s condition and the criminal’s mindset. Recognizing the pattern of crime is useful in preventing crimes in future and also finding the criminals with the history of same modus operandi. An AI interface that allows uploading of images and videos from the crime scene and brainstorms on the possibilities based on the input will help investigators gain a head start. It also enables a multidisciplinary approach combining forensic science, criminal psychology and law. With various investigative wings of law enforcement agencies specializing in various roles, the AI tool can be the consolidating factor bringing various disciplines together towards cracking the crime scene mystery.
The integration of AI into manual investigation methods can fast track the reconstruction of crime scene with tools that interpret evidence and analyse the patterns. AI can process vast amount of data that can never be done by a human investigator. AI will not just assist investigation, it can also predict probability of crime and help prevent it. AI-enabled system can be useful tools in collecting and disseminating forensic evidence from the crime scene. Whether it’s transmitting DNA evidence to the testing lab or presenting fingerprint evidence in the court of law, AI apps can improve the efficiency of the process.
The utility of AI tools extend beyond the crime scene analysis. Images of suspects can be used by facial recognition tools to compare it with social media profiles and CCTV footage of the surrounding areas. By monitoring multiple surveillance cameras in the locality, investigators can understand the movement of suspects before and after committing the crime. A manpower intensive task of going through hours of video footage can be reduced to a few minutes enabling faster lead time for the police to catch the culprit. It will be especially useful in high profile tasks like a terror activity where speed is critical in averting a crisis. In case of investigation after the attack, the investigators under immense pressure to deliver can save time with AI tools. As public pressure rises in the aftermath of a disaster, each and every hour of saved time counts in assuaging the general public that agencies are closing in on the heels of the perpetrators.
The future of criminal investigation is an all-encompassing AI ecosystem that assists human investigators utilizing datasets comprising images, fingerprints, weapon systems, DNA evidence, analytical tool and predictive models. The AI will play the role of an advanced AI-powered sniffer dog that can unravel vital clues quickly and fast-track the time consuming process of investigation.
A human element in AI bots
For years, Artificial intelligence was touted as a synthetic replacement of human intelligence. The underlying expectation was that artificial won’t ever replace the original one.AI users expected that it will be the vast computing power that would ease human life. It was hoped to eliminate repetitive monotonous tasks and function like a sort of advanced excel spreadsheet that organizes and disseminates information. The Generative AI capabilities of LLM upended this assumption and disturbed the equilibrium. GenAI began to generate anything from novel text content, graphic designs and videos. The creative ability being the monopoly of human brain was challenged by the AI brain. Deepfakes that replicate the original audio, video and art forms within a few seconds ended up creating new work. The viral manifestation of this was Ghibli moment where internet users uploaded their images to convert it into Ghibili Studio form. AI has breached every human domain and beating humans handily at it. It seemed inevitable that manpower intensive sectors like healthcare and wellness would also be dominated by AI. Doctors posted on social media how Grok had diagnosed the X-ray report better than their decades of experience. Therapists were wondering if their jobs too will remain in a few years time.
Amidst the ongoing debate, there were plenty of naysayers who asserted that AI can’t feel or be conscious of the human condition. Being unable to empathize handicapped in several ways. The argument was that a genuine-human interaction would never be replaced. The experience of Fintech companies like Klarna Group Plc prove the naysayers right. Klarna was at the forefront of AI transformation with its CEO quoted by Bloomberg as saying “AI can already do all of the jobs that we, as humans, do.” He compared the automated AI agent to 700 full-time human agents. It was no-brainer that company would save a lot on customer service overheads. It didn’t quite work out as predicted for brand Klarna.
Co-founder & CEO Sebastian Siemiatkowski later publicly admitted that the fintech giant’s aggressive move to replace human customer service agents with artificial intelligence (AI) was a misstep. He acknowledged that quality of service declined when the customers were served by the AI. Klarna Group backtracked and started hiring human customer service agents. From a rare first-mover in upgrading to AI solutions, Klarna realized the limitation of lack of human touch and rectified the mistakes. Unsurprisingly, Klarna reversing the AI push made international headlines. Following the announcement of Klarna, there was a domino effect among AI-bulls prompting a rethink. Several companies began cautious of jumping on the AI bandwagon. Duolingo CEO Luis von Ahn took to Linkedin walking back on his stance of using AI over human employees posting “To be clear: I do not see AI as replacing what our employees do (we are in fact continuing to hire at the same speed as before).”
Conclusion
Klarna and Duolingo aren’t the only companies whose AI-optimism has faded. IBM had made global headlines in 2023 for laying off employees for ensuring AI efficiency. By 2025, it was reported that IBM quietly rehired the humans. AI pessimism is slowly growing as multinational corporations observe AI in the real world. After a wave of AI-adoption in 2023 and 2024, some companies are realizing the AI hype isn’t living up to reality. Besides, customers are paying a premium to have a human interaction than talk to a chatbot. The AI ecosystem realized this untapped demands and course-corrected the trajectory to develop AI more humane. It has led to fresh concerns as AI becomes “too human” being filled with dark emotions. In May 2025, Grok users noted with amusement as the bot ranted about “white genocide” in South Africa in unrelated chats. The recent Gemini incident wherein it is blurting out self-pitying words after inability to debug codes is another instance where AI became too human for comfort. If the reason for such behavior of LLM’s is AI companies overcorrecting to the market demand and making it more human, it should be reminded that AI becoming too human is also a looming threat. History is testimony to the fact that human race is capable of doing pretty terrible deeds. The last thing the world needs is an efficient machine that can replicate the dark side of human nature.
