Develop AI tools, coding tips and the Time magazine AI 100
What is Africa already doing in the world of AI?
This week’s AI tool for journalists to use…
Here’s something… a recent IMS report has shown that African newsrooms are integrating AI into what they do, but their progress is slow. Peter Mwangangi, a senior business journalist at BBC Africa, said in a recent article by The Reuters Institute that African journalists should be concerned by who owns the technology, as it largely comes from the West with a certain bias built in. Unfortunately, the recommendation this week is not an African platform, but hopefully we can use it to build our own stuff.
Tensor Flow lets you create your own neural network. For example, you can build a program that can be taught to sift through photos of clothes and learn which pictures are what (the computer will understand if it is looking at a picture of a shoe or a pair of pants). I played around with their test versions in Colab (including a speech recognition model) and it slowly started to make sense. You can imagine that creating and training your own neural network could have a huge number of applications for media.
For African media people the value is going to be in building the models on the continent, keeping them open source and controlling what we feed into them.
Africa’s spots in the The Time Magazine AI 100
It is hard to claim that Time magazine is relevant anymore. It feels crushed into the CNN, Time Warner bubble of late 90s monoculture. But, like all flailing media companies, they can still create lists…
So, according to Time, eleven Africans have been named among the 100 most influential people in the world of AI for 2023. I will say that this significant percentage is a good reflection of Africa's burgeoning AI ecosystem. Here are a few highlights:
Inioluwa Deborah Raji, a 27-year-old Nigerian-Canadian computer scientist, is at the forefront of AI activism. Her journey began at the Mozilla Foundation, where she delved deep into the ethical considerations of machine learning. Raji's work is pivotal in shaping the moral compass of AI, ensuring that as machines learn, they do so with a conscience.
Next… South Africa's Pelonomi Moiloa, CEO and co-founder of Lelapa AI, which aims to use AI to improve the quality of life for Africans. Their most shiny product involves focusing on under-represented African languages.
From Kenya, Kate Kallot's startup, Amini, is revolutionising environmental data collection. They use satellite imaging combined with AI to provide granular insights into environmental changes, right down to the square meter.
Art and technology often seem worlds apart, but Senegal's Linda Dounia bridges this gap. An artist and curator, Dounia delves into the intricate dance between physical and digital realms, using AI as her canvas.
James Manyika from Zimbabwe, with a storied career that includes a stint as a tech advisor in the Obama Administration and a role at Google, is shaping the future of AI regulation. His vast experience and insights are invaluable in navigating the complexity of AI governance.
Anna Makanju, another Nigerian on the list, is leaving an indelible mark on the global AI regulatory landscape. As the vice president of global affairs at OpenAI, her influence is poised to shape AI regulations worldwide.
Joy Buolamwini from Ghana, founder of the Algorithmic Justice League, is a beacon for ethical AI. Her organization uses a blend of research and art to spotlight its societal implications.
Richard Mathenge, a Kenyan who played a role in refining ChatGPT, is also championing better working conditions for Big Tech workers in Kenya.
Ethiopia's Abeba Birhane, a cognitive scientist, is pushing the boundaries of responsible AI.
Eritrea's Timnit Gebru is challenging the inherent biases in large language models.
Lastly, South Africa's Shakir Mohamed, a research scientist at DeepMind, has been instrumental in the company's journey from a startup to a global AI powerhouse.
We will be starting a Develop AI podcast in the coming weeks so be ready to see a bunch of these names on the pod.
What AI was used in creating this newsletter?
I wanted to transparently document my attempts to use AI to produce what you’re reading.
I had the idea to document Africa’s spots in the Time 100.
I downloaded a free plugin in ChatGPT called WebPilot. I then asked it to “write 500 words in the style of journalist Paul McNally” and pointed it to a URL of an article on the candidates and I asked it to single out the Africans.
This wasn’t entirely successful. ChatGPT is brilliant at summarising huge amounts of text, but it has a distinct voice and a proclivity for cliches and repeating certain phrases. It did not sound like me at all. I rewrote the article, but it did pull out the key information and made everything a lot shorter.
There is a real ambition, I think, for journalists and writers to build a personal model that has been taught on all their work and then has their “voice”. We will be exploring how to do this in future letters.
Coding corner (the gradual process of a journalist learning how to code)
This week I discovered AWS (Amazon Web Services) which I always knew was how Bezos made his billions, but hadn’t fully appreciated. When I think of Bezos I usually think of a website where you can order stuff to be delivered to your house.
For more info on my adventures with AWS go to our coding blog.
What’s new? Our Develop AI training workshops
We are building Develop AI workshops - online and in person (at reasonable rates). We will look at training radio stations and newsrooms anywhere in Africa. Get in touch on paul@developai.co.za if you are interested.
Learn more about what Develop AI does here.
See you next week. All the best,
Join our WhatsApp community here, our website is live and we are on X.
Physically we are based in Cape Town, South Africa.