Top data challenges that keep AI professionals in Africa up at night

We then jumped into the main event of breaking out into rooms to exchange solutions to common challenges. I moderated the session discussing challenges around data and describe it in the next section. 

In the following section, I will tell you about the “challenges around Data” room I moderated. But feel free to read my colleague’s session on Jobs in Africa, or even sign up to volunteer with us as we make history! 

Data and all its challenges, from realism to access.

When it comes to all things AI-related, we require data. It is the foundation of any learning and algorithm we create and all AI practitioners will spend the majority of their time dealing with the data to ensure their quality and quantity in terms of accuracy, representation, feature selection, etc.

Thus it comes as no surprise that we face significant challenges in the sphere of data in a developing continent such as Africa. 

Top 2 Challenges

Lack of data that reflects real-life standards and processes:

The Data used for training of data scientists/ AI engineers should be more realistic to real-world standards from the collection process to the engineering ~ Brought up by Rose Delilah Gesicho, Nairobi Women in Machine Learning – Kenya

Solutions

1. We should implement Practical Training programs that use real-life data and process ~ Janet Akorfa, Yielding Accomplished African Women – Ghana
2.  Setting up guides for dealing with real-world data through the experience the communities have had when collecting the data through the documentation of the processes we have been through ~Janet Akorfa, Yielding Accomplished African Women – Ghana

Lack of publicly available/open-source data and the associated costs of collecting our own data ~ Brought up by Perez Ogayo, ALU AI-LAB – Rwanda

Solutions

1. Partnerships with companies to solve their challenges and use their data (create internships for students) ~Janet Akorfa, Yielding Accomplished African Women – Ghana
2.  Ensure we build a network with the right companies to make the process easier and partner with them ~Isaac Edem Ayitey, Cheetahs Ltd – Ghana
3.  Partner with organizations and companies to aid their data collection. Boots on the ground style for fresh and accurate data (with an agreement that community gets to use it in their programs and training) ~Fabian Fawole, Alliance4ai, Nigeria

Below are more challenges that were brought up that will be discussed informally due to time constraints in the meet:

• How to handle large datasets for big problems
• Expensive to get the data we need for the problems we want to solve
• How to generate our own data when there’s no public one

Do you have answers to these challenges or think there are other challenges beyond the ones identified by our attendees? Complete this form to tell us more and attend the next gathering if you lead an AI community!

Numerous solutions exist to help solve the problems highlighted above and we believe this initiative is a step in the right direction to help African AI communities to break the current status quo. Continue the conversation by tagging us on twitter (@alliance4ai) or posting on our Forums module. 

If you want to join us in shaping this great initiative, tell us more about yourself here.

Have a blessed week!