How an AI bot cracked the tough NEET exam, beating most human scores

The results of India's competitive medical entrance examination — the National Eligibility cum Entrance Test (NEET) 2025 — have recently been announced, but it wasn't just human aspirants in the race this year. As students tackled the high-stakes paper, Allen, one of the country's leading coaching institutes, quite literally put its AI-powered bot 'Allie' to the test.
As a result, while Allen's top-performing student, Tanmay Jagga, secured an All India Rank (AIR) of 74 with 650 out of 720 marks, Allie outshone him by some margin with 678 marks, a score likely to fetch a candidate a top-10 rank and a seat at the prestigious All India Institute of Medical Sciences (AIIMS), Delhi, according to the coaching institute.
Thankfully, the bot has not been developed to directly compete with students but to build trust in the AI-driven complimentary support system offered to students for academic problem-solving. "We are pedagogically and result-wise quite sound but there was always a scope for improvement. We realised that apart from in-class learning, a lot happens outside of class where students practice, have doubts, and get stuck," said Ankit Khurana, Allen's chief product officer on the company's motivation behind the bot. "In this journey, students face several challenges but teachers can't be a support for every student. So, we thought our digital platform could be a companion for students. With that philosophy, we created 'Allie'."
According to data shared by the institute's chief technology officer Saurabh Tandon, over 90,000 students used the bot to solve nearly 600,000 queries last month. Out of these, roughly 80 per cent of doubts were immediately responded to while the rest were transferred to a human teacher for resolution.
Allie, who somewhat resembles ChatGPT in appearance, answers the questions within seconds but for complex questions, such as those comprising images, it can take up to three minutes, Tandon said. The bot was initially launched in December 2024 and was trained for biology subjects but by March 2025, it covered almost the entire syllabus for the NEET examination, barring some topics like organic chemistry. That omission could explain why Allie got two questions wrong and didn't attempt eight questions in the NEET exam.
As for the underlying technology, the company said it relies on a proprietary verifier model and a dynamic routing system. While a proprietary verifier model refers to the use of targeted strategies to verify answers by cross-referencing them with curated study content, including theory and Q&A (question and answers) archives, the dynamic routing system sends queries to a suitable language model based on the problem type (image-based, text-based, or reasoning-based question), and other metrics.
Going forward, the company plans to train the bot with the NEET examination syllabus in its entirety while also assisting students preparing for other examinations such as the Joint Entrance Examination (JEE) for admission to undergraduate engineering programs. In the ecosystem of Allen, around 5 lakh students prepare for JEE and NEET examinations as paid subscribers.
"The interaction with Allie has grown roughly six times in the last six months, that's something which we did not expect. The amount of interaction that happened was unprecedented. As we improve the technology, we will do the same for JEE examination. We will also keep on refining the bot for NEET in terms of the quality of answers and conversation," Khurana said.

Try Our AI Features

Explore what Daily8 AI can do for you:

Learn with AIFact CheckingText to SpeechAI Summary

Related Articles

NDTV

4 hours ago

• NDTV

West Bengal Joint Entrance Examination 2025 Result To Be OUT Soon, Details Here

West Bengal JEE Result 2025 : The West Bengal Joint Entrance Examinations Board (WBJEEB) is expected to declare the result for West Bengal Joint Entrance Examinations (WBJEE) 2025 soon. Once released, candidates will be able to download the result on the official website, WBJEE-2025 was held on April 27, in two shifts. The first shift conducted Paper I (Mathematics) from 11 AM to 1 PM, while the second shift was held for Paper II (Physics and Chemistry) from 2 PM to 4 PM. West Bengal JEE Result 2025: How To Download JEE Result ? Visit the official website, Under the "Important Link" section, click on "Rank Card For WBJEE 2025". Enter your application number, date of birth and security pin. Click on "Sign in" button. Your result will be displayed on the screen. Download your result for future reference. Candidates must make sure to cross-check the identification details on the marksheet such as Candidate name, Roll number, Date of birth, Category and Gender. Also, the result specific details on the marksheet include: Rank secured in WBJEE 2025 Total scores obtained Individual scores in Physics, Chemistry & Mathematics. The WBJEE 2024 exam was held on April 28, 2024, and the results were declared on June 6, 2024. The provisional key for WBJEE was released on May 9, 2025.

Time of India

7 hours ago

• Time of India

India must boost production capacity for military-grade drones: Defence Secretary

Live Events (You can now subscribe to our (You can now subscribe to our Economic Times WhatsApp channel Highlighting the role of Unmanned Aerial Vehicles (UAVs) in modern warfare Defence Secretary Rajesh Kumar Singh has said that India must boost its production capacity for military-grade drones. He noted that the Centre wants to cut the defence procurement cycle to a maximum of two years and the government will come up with a new Defence Acquisition Procedure DAP 2025) in six months."I would say that India has a significant manufacturing ecosystem for drones, but a lot of it is really for civilian drones. Military-grade drones, not so much. There are, I would say, three to four, five manufacturers who have some capacity in that area. It's something we'll have to build up quickly," Singh told in an interview."And to the extent that, like all military purchases, it's a monopsony. The government is the only buyer. And I guess the government will have to use its buying power to provide the resources and give visibility in terms of orders to that sort of small ecosystem, which hopefully will gradually become a large ecosystem of military-grade drone manufacturers," he said that military-grade drone production in the country needs some transfer of technology, as a lot of the technology may not be available domestically."I would say that we are looking at least three to four manufacturers who are coming up with this type of capacity in terms of military grade drones. We will encourage them through orders and through tenders so that they can start investing in this area even more, including tying up with a lot of the technologies from outside," he said."The volume that we can generate in terms of the size of the orders that can be bigger than the combined turnover of all the other civilian drone manufacturers in the country. That type of ensuring that they have a full order book if they make investments for inter-military grade drones, we will try to ensure by speeding up procurement and by giving them visibility in terms of orders," he Defence Secretary said he would like to see a situation where the maximum time taken for any procurement, even the most complex ones, should not exceed about two years."Defence Acquisition Procedure 2020 is a very bulky book. It's a bible for the defence industry. We need to make it simpler. A lot of redundant steps can be avoided. A lot of things can be done concurrently rather than sequentially. And a lot of simplification in terms of the kind of companies that can bid, where we've put in some unnecessary conditionalities, all those have to be removed," he said."So, for that, we have a committee now working. The target date is six months from today. So you'll have a new DAP. It will be DAP-25 rather than DAP-20," he added.

Time of India

7 hours ago

• Time of India

Social media ban moves closer in Australia after tech trial

HighlightsAustralia's world-first social media ban for users under the age of 16 is set to be implemented following a successful trial demonstrating that age verification technology is both feasible and can be integrated into existing platforms. The government-commissioned Age Assurance Technology Trial concluded that there are no significant technological barriers to preventing users under 16 from accessing social media accounts, with various methods tested including facial scans and behavioral inference. Digital platforms such as Snapchat, Meta Platforms Inc.'s Instagram, and X will be tasked with enforcing the new age limit, facing penalties of up to A$50 million ($32 million) for non-compliance. Australia's world-first social media ban for under-16s moved closer to implementation after a key trial found that checking a user's age is technologically possible and can be integrated into existing services. The conclusions are a blow to Facebook-owner Meta Platforms Inc., TikTok and Snap Inc., which opposed the controversial legislation. Some platform operators had questioned whether a user's age could be reliably established using current technology. The results of the government-backed trial clear the way for the law to come into force by the end of the year. The findings also potentially allow other jurisdictions to follow Australia's lead as countries around the world grapple with ways to protect children from harmful content online. 'Age assurance can be done in Australia and can be private, robust and effective,' the government-commissioned Age Assurance Technology Trial said in a statement Friday announcing its preliminary findings. The trial's project director, Tony Allen, said there were 'no significant technological barriers' to stopping under-16s gaining social media accounts. 'These solutions are technically feasible, can be integrated flexibly into existing services and can support the safety and rights of children online,' he said. Under the new law, digital platforms including Snapchat, Meta-owned Instagram , and X will be responsible for enforcing the age limit, with penalties of as much as A$50 million ($32 million) for breaches. The trial tested a range of methods and technologies, including facial scans, inferring a user's age based on their behavior, age verification, as well as parental controls. The tests also took into account the ways teenagers might try to circumvent age checks. 'We found a plethora of approaches that fit different use cases in different ways, but we didn't find a single ubiquitous solution that would suit all use cases,' the trial said in its statement. More than 50 companies participated in the trial, while Apple Inc. and Google, developers of the most popular mobile-phone operating systems, are also contributing, Allen said on a video conference call on Friday. The trial didn't assess public acceptance for any particular technology or the costs involved. The accuracy of the different methods, for example the margin of error for facial analysis, wasn't made available.