As we look ahead to 2025, there’s a reality we can’t ignore: AI is transforming every industry, every aspect of our lives. But here’s something that often gets missed in the conversation—AI, just like the internet, doesn’t stop at any borders. It’s global, always on, and constantly learning. But it’s also a product of the people who build it, which means it can come with built-in biases that can affect different communities in very real ways.
Let’s get real: If AI is trained primarily on data from one part of the world, how can it truly understand the experiences of people from another? This is something I’ve seen firsthand as an African American who’s been in tech for decades. AI systems often don’t see me—or people like me—the way they should. Whether it’s voice recognition software that struggles with accents, or facial recognition systems that misidentify darker skin tones, the biases in AI are real and they affect people’s lives.
The Global Nature of AI—But Countries Aren’t the Same
The internet is borderless, and that’s been its greatest strength. AI, built on the back of this digital world, inherits that same global reach. You’ve got AI models developed in Silicon Valley, trained on datasets from Europe, and then deployed in places like Africa and Asia—all in real-time. But here’s where it gets tricky: what works in one country might not fly in another, especially when those AI systems haven’t been trained with cultural sensitivity in mind.
Take it from someone who knows—being African American in tech is a constant reminder of how convenience AI can miss the mark. I can’t tell you how many times voice assistants couldn’t understand my requests or how often AI-generated content seemed tone-deaf to the cultural context of Black communities. Now, imagine how these biases get amplified when AI is scaled globally, affecting millions, even billions, of people who have different languages, cultures, and ways of thinking.
Why We Need Global Standards—But Not Just One Size Fits All
AI has the power to do a lot of good, but it can also cause real harm if it’s not done right. That’s why there’s been a lot of talk lately about creating a global code of practice for AI. The goal is to make sure AI is transparent, ethical, and doesn’t end up causing harm. But here’s the thing: if we make these rules too rigid, they could slow down innovation, especially for smaller companies or startups that don’t have the resources to keep up. At the same time, if we leave it too loose, we risk letting biases slip through the cracks.
And when it comes to biases, let’s not sugarcoat it—AI models can perpetuate stereotypes and inequalities if they’re not developed with diverse data. I’ve seen it firsthand. From healthcare algorithms that misdiagnose conditions in Black patients to hiring systems that overlook qualified minority candidates, the risks are too real to ignore.
The Tipping Point: Why 2025 is the Year of Decision
In 2025, AI adoption is set to hit levels we’ve never seen before. It’s going to be in your car, your doctor’s office, your bank, and even your pocket. But without clear rules in place, we’re headed for chaos. Imagine a world where different countries have conflicting AI regulations—that’s going to slow down progress and create a “splinternet,” where the digital world gets carved up into disconnected pieces.
To avoid that mess, we need to get serious about three things:
- A Global Framework with Flexibility:
- Local Adaptation:
- Collaborative Enforcement:
So, What’s Next?
We’ve got a draft out there now—a General-Purpose AI Code of Practice. It’s not perfect, but it’s a start. The goal is to get feedback from everyone—governments, companies, and everyday people like you—before it becomes the real deal.
The key takeaway here? AI is too powerful to leave unregulated, but it’s also too global to be governed by any one country alone. We need to find that balance, and we need to do it fast. And as someone who’s been in this tech game for a while, I’m telling you—if we don’t fix these biases, we’re going to end up with technology that doesn’t serve everyone equally.
For those interested in reading the current draft and adding your voice to the conversation, check it out here: [link to the AI Code of Practice].