If you’ve recently read a story on Daily Graphic, GhanaWeb or Ventureburn about a 19 year old Ghanaian boy developing an AI app for the detection of plant diseases, with 30,000 farmers as users and then winning $600,000 for it, you might want to know some truth behind the smokescreen. In short, the app is a rip-off of karaAgro AI, a project Mustapha Diyaol-Haqq worked on as an intern under GUDRA. Also, they are more than likely lying about having reached 30,000 farmers without the app being on any app store or distribution platform, that’s very much ridiculous. Okay then, I’m going to share some details of this story as well as some screenshots to serve as evidence.
Lets start from the 6th of August 2018, when Mustapha reached out to me and asked if I could be his mentor as he is passionate about learning Artificial Intelligence (AI) and working with me at GUDRA/minoHealth (check Image 1). I agreed and sent him some materials and would periodically make time to help him with AI, and sometimes even his personal challenges.
Mustapha then attempted working on a breast cancer AI project (a standard project and dataset used by beginners), as proof he’s capable of joining our team and working with us on our AI systems. He asked for my mentorship on his write-up on this project, which I offered. (check Image 2)
On November 20th, 2018, I reached out to Mustapha, asking if he would like to intern with us and work on the android development aspect of a project we have been working on, karaAgro AI. He agreed, and became an intern. (check Image 3) This app is an AI-powered plant disease detection app. We had trained the Deep Learning (AI) systems in the detection of 23 diseases from about 10 crops, using publicly available data. The models were specifically Convolutional Neural Networks. Mustapha did not contribute to the development of these AI systems, only the Android Development. To be able to contribute to the app development, Mustapha was given access to the AI systems and data.
After weeks and months of Mustapha not delivering any tangible and worthwhile results, whilst repeatedly disappearing, being unreachable and coming back with endless excuses, we thought it best to take him off the project but not terminate the internship. (check Image 4) As matter of fact, one of my colleagues advised we limit his access to confidential materials of GUDRA, because of how untrustworthy he had become. My colleague suspected some foul play. The few java code Mustapha contributed turned out to be so bad we had to remove it all and redo everything. (check Image 5)
In June 2019, before we released our app, we then found a website (www.okuafofoundation.org) emerging and claiming to have an app doing the same thing karaAgro does. Now, we thought this was simply an alert for us to speed up development since competition is already here. We found out later the person behind this project is none other than my mentee, Mustapha Diyaol-Haqq and Tina Appiah, founder of Ghana Code Club. We then attempted to handle this privately, we called Mustapha several times which he refused to answer. We then reached out to Tina and told her their project is the very project Mustapha was invited to work as intern. She claimed to have not known this and was only aware he interned with us. She then promised to have a conversation with him and have him call us so we can resolve this peacefully. We later received a call from Mustapha, the first thing he starts talking about is the internship agreement he signed and how it does not prevent him from implementing a project of such nature. He then asked if we can have a meeting to analyse the agreement. Note, this is a 19 year old boy. He wasn’t denying ripping off a project he was invited to contribute to, an unreleased project of his mentor he claimed he was honoured to have. He was interested in legal bindings and loopholes. I then informed him we are not interested in taking legal actions against them, I wanted to know if first of all, he thought this was right. And we then told him the action we would take is to expose them publicly. The next day, Tina reached out and said they have discontinued the project as she knows how painful it can be to have a project ripped off.
We later released the beta version of karaAgro AI, August 12 2019. We have since been working with plant pathologists, researchers and experts in University of Ghana Crop Science Department and West African Centre for Crop Improvement (WACCI) piloting and improving the platform. Developing an AI system that can truly generalize and detect disease in various crops across various farm settings, light conditions and mobile devices is a tough problem. Now, we are not the first to attempt working on this problem. It’s a hard problem. What is easy, apparently, is pretending to have done the work when you haven’t. Okuafo Foundation likely doesn’t even have a fully functioning app yet. If they do, they should place it on the app store so it can be verified. Beyond a functioning app, they more than likely haven’t solved the problem of building an AI system that can generalize properly across conditions.
Even further, claiming to have reached 30,000 farmers in a period of just a few months, without the use of an app distribution platform is silly to say the least. The further claim of having helped farmers to increase their crop productivity and harvest by 50 percent is very much questionable. It’d be very unfortunate if the Zayed Sustainability Prize awarded this project $600,000 without verifying such claims they made. It’s also sad reputable media outlets, including Daily Graphic would write about this project and their claims without much due diligence. This is however not uncommon in the tech sector and the media sub-par coverage sometimes.
Now, this article is not a claim of an IP infringement, as no individual entity can claim ownership of AI applications in Agriculture and plant disease detection. Neither do I have a problem with competition, even if it’s from an old intern or employee. It is not uncommon for people to leave tech companies and become a competitor. The problem here was the conditions under which they ripped off an unreleased project and created a website claiming to have developed it. Another problem here is making bold claims that are more than likely false, using such claims to raise funds and such claims being shared by media outlets without much verification. This is almost a mini-case of another Theranos. It is not uncommon in the Technology sector for characters and companies like this to emerge, that is true for Ghana as it is for the USA and its Silicon Valley. However, it’s very important that with each case, the truth is sought after!