Democratizing AI
There has been lots of research and developments in the field of Artificial Intelligence and Machine learning, with lots of billions of dollars being invested by corporate organizations and research institutions. The outcomes and discoveries of these research works are mostly implemented by top companies, government entities and few individuals with the funds to afford the compute resources and technical team to harness the possibilities in this field. Although most people have the opportunity to experience AI and machine learning driven technologies such as search engines, online recommendations, chatbots, keyboard auto-correct, voice assistants, online APIs and more, these technologies do not provide people the choice to decide how, when and at what scale Artificial Intelligence can be integrated with their social and economic lives.
Existing AI platforms accessible to the public have the following Challenges:
1. LACK OF CUSTOMIZATION : Current AI technologies provided are made to satisfy the average requirements of most people, thereby denying them the opportunity to customize the use of AI technologies for their individual needs.
2. LACK OF EXTENSIBILITY :
Current AI technologies provided publicly lack extensibility, making it impossible for users to improve their AI implementations to satisfy new conditions and demands in their social and economic lives.
3. CONNECTIVITY PROBLEMS :
Due to the high computational requirements that AI and machine learning has, most people have to depend on consistent connectivity to cloud services to use these technologies. This leads to high bandwidth cost when a heavy workload is involved, and it turns out to be expensive using Cloud services on the long-term basis.
4. ALGORITHMIC BIAS: A number of AI technologies and machine learning tools were built and trained on data that fit specific race, economic, cultural and social environment. This leads to poor performance and sometimes unpleasant results when attempts are made to utilize these technologies elsewhere. An example is that of facial recognition problems discussed by Joy Buolamwini at TEDxBeaconStreet .
5. TECHNICAL PROBLEMS : To harness current state-of-the-art technologies in AI and Machine Learning requires highly technical knowledge and skills in prgramming and related mathematics, but very few people with such capabilities are available.
I and my brother John Olafenwa are determined to address these challenges and ensure the future of AI is that in which everyone have control of basic AI capabilities that are essential to business development, economic growth and better social well being.
The project Democratizing AI is aimed at developing technologies that will provide AI and Machine Learning tools, libraries and hardware that will allow businesses and individuals to build, manage and maintain their AI resources conveniently without issues of technicalities, connectivity and bias. They include and not limited to the following:
TOOLS & LIBARIES FOR DEVELOPERS
We will be releasing a couple of tools and libraries that will abstract the low level operations in AI as well as Machine Learning processes. These tools will allow developers to build applications and systems that ship inbuilt AI capabilities without connectivity issues, especially in remote locations where AI is needed but connectivity is minimal or nonexistent. They will also allow for easy development of extensible AI applications.
TOOLS FOR NON-DEVELOPERS
There are lots of individuals and business who can benefit from state-of-the-art AI technologies but does not have the technical know-how. We will be releasing user friendly Applications that will allow individuals and businesses to customize, integrate and extend AI and Machine Learning resources.
SPECIFIC HARDWARE
State-of-the-art AI technologies, especially Machine Learning requires highly powerful hardware to perform properly, especially in real-time applications. Most computer and smartphone users do not have this kind of hardware at their disposal. We will be working on a separate and independent hardware that can be comfortably integrated with any computer enabled system like smartphones, personal computers, drones, smart cameras, ATMs, etc .
