Alan has been on a mission to make healthcare more accessible since 2016. We recently reached a new milestone with Mo, our AI-powered medical assistant, deployed within Alan's medical advice chat.
Below are the results of the first large-scale study examining conversational medical AI in real-world conditions.
Everyday, we ask Google more than 1 billion health-related questions. This number highlights a gap: we want to understand and take action for our health, but we lack reliable and accessible guidance. This is the challenge that Mo aims to address.
Mo is an AI-powered medical assistant, designed to support you with Alan's medical expertise every day.
To continuously evaluate and improve Mo, we integrated it into our existing medical chat service, where Alan members regularly communicate with doctors. Over a three-week period, we included more than 1,500 conversations in the study, of which 926 were eligible for interaction with Mo. We offered Mo to half of the eligible patients to compare patient experience with and without Mo.
Mo's development follows a responsible approach to healthcare innovation. Every message is verified by a doctor within 15 minutes, ensuring Mo’s accuracy and relevance. We implemented essential safeguards: data hosting within the European Union, adherence to the strictest security standards, clear differentiation between AI and doctors in the interface, and explicit patient consent for AI use.
As AI in healthcare often raises concerns, we wanted to understand how patients would respond to Mo. The results were compelling: in 81% of conversations, patients chose to interact with Mo, even though they had initially come to speak with a doctor. This adoption rate suggests that patients are ready to embrace AI in their healthcare journey - just as they've already integrated the internet into their health practices today.
We also observed strong engagement in conversations where Mo was present, thanks to its instantaneous responses. Conversations were more dynamic, with patients responding 2 to 3 times faster to Mo than to doctors (57% of responses to Mo came within one minute, versus 27% with doctors).
For example, thanks to instant responses, rather than switching to something else and waiting for a notification, users continue their conversations with Mo without putting down their phones. Our hypothesis? This engagement facilitates information exchange, resulting in better guidance.
Patient satisfaction is key to AI adoption. Mo's results in this area are remarkable: conversations assisted by Mo received an average rating of 4.6/5, slightly exceeding the control group conversations. Patients also rated the clarity of information higher. More notably, patients reported equivalent levels of perceived empathy and trust compared to traditional conversations, even while knowing they were interacting with AI.
Despite substantial volumes, we observed no significant disparities in Mo's acceptance across genders or ages. In other words: Mo appears to be a universal solution. Men, women, young people, seniors: all groups are equally represented.
Doctor collaboration was integral to Mo's development. Our Product team included a doctor to ensure Mo's design would serve medical professionals' needs and reliably serve patients.
When we presented Mo to our medical chat team - 18 doctors who had been with Alan for over two years on average - we were pleased to be met with curiosity where we had expected skepticism.
The results validate this collaborative approach: doctors rated 95% of Mo's conversations as "good" or "excellent". No conversation raised patient safety concerns, thanks to their supervision.
"I find Mo very effective and believe it fully embraces its role as an assistant," confirms one of the general practitioners working on the medical chat.
The detailed results of this study are available here.
These initial results are encouraging, but they're just a first step. Our vision for Mo goes further: we want to make it a health companion who supports you long-term. An assistant that answers your questions, understands your context, and bridges the gap between you and healthcare professionals.
Artificial intelligence, used ethically and transparently, can transform healthcare for everyone's benefit. With Mo, we're laying the first foundations for a more accessible, preventive, and human healthcare system.
We still have (plenty of) work ahead to provide increasingly accurate and reliable answers to patients. If you're passionate about AI and building better healthcare, apply.
A huge thank you to all of the general practitioners at the Alan Clinic who supported Mo's development with their goodwill and involvement. None of this would be possible without them!