Project summary

AI safety's most important ideas live in papers, technical blog posts and long-form essays that often assume a machine learning background. This makes it difficult for people without a technical background to engage with the field.

As AI capabilities advance, this gap could become wider. People who will build, govern, work with or be affected by advanced AI systems need ways to understand the ideas shaping the field.

CoLoG is starting with a free, three-week AI Safety x Content Fellowship for 30 Africans. Fellows will learn the core ideas of AI safety and turn important research, ideas and catastrophe scenarios into short videos that the public can understand.

The first cohort has already received 300+ applications within 12 days of the announcement.

What are this project's goals? How will you achieve them?

The immediate goal is to help 30 Africans understand AI safety while creating useful content that makes the field easier for other people to discover.

Over three weeks, fellows will learn how AI systems work, explore AI capabilities and risks, engage with major AI safety ideas, and create an explainer video based on something they have learned that they want the public to know.

The longer-term idea is to establish AI safety reading groups at leading universities across West Africa and embed the learn-and-create model into them.

AI safety research is growing quickly. We need a way to keep making new knowledge understandable as the field develops. Embedding content creation into university communities gives us a way to do this while also helping more students discover AI safety.

The fellowship is therefore the first step towards building a model that can grow across universities, expand the public knowledge base and identify people who want to go deeper into AI safety.

How will this funding be used?

Minimum: $2,000

Participant data support: Internet data packages for 30 fellows so they can attend the virtual fellowship, access readings, use AI tools and submit their work: $1,000

Video creation credits: AI/video production tools that fellows will use to create their explainers: $600

Core programme infrastructure: Zoom, Slack, hosting, storage and other essential tools: $400

Ideal: $5,000

Participant data support: Internet data packages for 30 fellows: $1,000

Video creation and AI credits: Tools for research, scripting, voice generation, video creation and editing: $2,000

Programme infrastructure: Zoom, Slack, website/hosting, storage and other tools needed to run the fellowship: $1,000

Content publishing and distribution: Tools and services to publish, organise and share the resulting AI safety explainers publicly: $750

Contingency: A small buffer for unexpected programme costs: $250

Max = $15000

Ideal budget x 3 cohorts (Sep, Oct, Nov)

Who is on your team? What's your track record on similar projects?

I am the founder of CoLoG and have experience building communities and learning programmes at scale. I led a climate change learning programme across five leading universities in Ghana, coordinating 200+ campus ambassadors who reached more than 10,000 young people. I also helped establish the Ghana Nonprofit Professionals Network.

Through my involvement with PauseAI, including grants prospecting, I have gained practical exposure to AI safety communities and how initiatives in the field are developed and supported. I also recently completed BlueDot Impact's AGI Strategy course.

Our mentors bring additional experience from the AI safety and communications fields. Oscar Mata, Head of Design at FAR.AI, will mentor fellows on translating complex AI safety research into clear and engaging content for public audiences. Other mentors bring experience across AI safety, design and communications.

What are the most likely causes and outcomes if this project fails?

The biggest risk is that fellows learn the material but struggle to turn difficult AI safety ideas into content that people without a technical background can understand.

We will address this through practical assignments and feedback from mentors throughout the content creation process.

Another risk is that some participants cannot fully take part because of internet costs or limited access to video and AI tools. This is why data support and creation credits are a core part of the budget.

If the project succeeds, we expect to have 30 people with a stronger understanding of AI safety and a first collection of accessible AI safety explainers that can be shared publicly

How much money have you raised in the last 12 months, and from where?

CoLoG is an early-stage initiative and this is our first major fundraising effort for the fellowship.

So far, I have primarily invested my own time and resources into developing the programme, including the curriculum, website, recruitment and programme design.