At first, John Pasmore was intrigued by ChatGPT.
The serial founder has been working in the field of artificial intelligence since at least 2008. He recalled a time when experts declared that it would be decades before something like ChatGPT appeared in the world. Fast forward – now the day has arrived.
But there's a problem.
ChatGPT, one of the world's most powerful artificial intelligence tools, suffers from cultural differences. This is very frustrating for black people like Pasmore. In fact, this oversight has drawn the ire of many black people who already don't see themselves adequately represented in the algorithms that they tout will one day save the world. Currently, ChatGPT provides too generalized answers to specific questions tailored to specific communities, as education appears to have a Eurocentric and Western bias. This is not unique. Most AI models are not built with people of color in mind. But many Black founders are adamant that they will not be left behind.
Numerous Black-owned chatbots and versions of ChatGPT have emerged in the past year to specifically cater to Black and brown communities, as Black founders like Pasmore seek to capitalize on OpenAI's cultural shift.
“Typically, if you ask a model who the most important artists in our culture are, the names that come up are Leonardo da Vinci and Michelangelo,” Pasmore said of ChatGPT. “You won’t say anything about India or China or Africa or even African Americans, because there is a bias that focuses on the historical trajectory of Europe.”
So Pasmore launched Latimer.AI, a language model that provides personalized answers that reflect the experiences of black and brown people. Erin Reddick started ChatBlackGPT, a chatbot centered around black and brown communities. There is Canada-based Spark Plug, a ChatGPT for black and brown students globally. Africa is also witnessing massive innovation in this field, with language models emerging to support the more than 2,000 languages and dialects spoken on the continent, which are still overlooked by Western AI models.
“We are the keepers of our own stories and experiences,” Tamar Huggins, founder of Spark Plug, told TechCrunch. “We need to build systems and infrastructure that we own and control so that our data stays with us.”
Personalized AI is here
Generalized AI models cannot easily capture the African American experience because many aspects of the culture are not online. Currently, algorithms source from the internet, but many traditions and dialects of African American culture are passed down orally or in person, so there are differences in the nuances of what the AI model understands about the community and what is actually happening.
That's one reason why Pasmore built Latimer.AI with a focus on accuracy rather than training on user-generated data scraped from the Internet, trying to use sources like the Amsterdam News, one of the oldest black newspapers in the country. With this, he began to see the differences between his model and ChatGPT.
He once asked ChatGPT about the Underground Railroad, a passageway that black Americans used to travel to northern states to escape slavery. ChatGPT's model refers to runaway slaves, while Latimer.AI adjusted the wording by referring to “slaves” or “freedom seekers.” This is more consistent with the more socially adapted we are discussing formerly enslaved people.
“There are some nuances to the language the model uses because of the training data, and the model itself only thinks about black and brown people,” Pasmore said.
Meanwhile, Erin Reddick's ChatBlackGPT is still in beta mode, with plans to launch on June 1st. Her products literally work. This is a chatbot where you can ask questions about black culture and receive personalized responses. “At the core of what we do is truly community-driven,” she said.
She is in the process of building a tool and is asking users how it should look and work. She also partners with educational institutions, such as Historically Black Colleges and Universities (HBCUs), to work with students to help teach and train her algorithms. “I want to create balanced learning opportunities so that black and brown people have a safe space to explore AI,” she said.
“The algorithm prioritizes Black sources of information, allowing us to speak to a body of knowledge that is more immediately relevant than the typical experience,” she told TechCrunch, adding that, like Pasmore's product, it's technically available to anyone.
Tamar Huggins built Spark Plug to provide a more personalized experience for black and brown communities. Her platform translates her teaching materials into African American Vernacular English (AAVE), an ethnic language associated with the black community. The dialect is traditionally transmitted orally rather than learned and written down like standard English. This means that the accuracy of AI models (or people) learning them over the Internet will be less accurate. Accurately capturing AAVE is important so that the chatbot can use it to respond, as well as make it easier for students to write prompts that will allow the AI to return the results they need.
“Creating content that resonates with Black students lets them know they are engaged in their education, which is critical to high engagement and academic success,” Huggins said. “Given the opportunity, Big Tech almost always puts profits ahead of people. So we created our own lane within the AI space.”
To capture the essence of AAVE, Huggins trained the algorithm on the speech of black writers from the Harlem Renaissance, black writers in education, and even his teenage daughter. Huggins also works with educators, linguists, and cultural experts to review and validate Spark Plug's output.
Pasmore is also working to expand the company into schools, especially HBCUs, as more students use ChatGPT every day to get their work done.
“This is a better AI partner for a lot of the work that black and brown kids need to do,” he said.
Uniting the diaspora
Africa is currently overlooked in the AI movement. For example, according to the 2023 Artificial Intelligence Index report, only 0.77% of global AI journals originate from sub-Saharan Africa, compared to 47.1% and 11.6% in East Asia and North America, respectively. Compared to North America in terms of population, Africa accounts for about 17% of the world's population, while North America accounts for only 7%. When the time comes to draw on information and expertise on AI, it is very unlikely that sub-Saharan research will be used, which could have implications for the development of global AI tools.
Africa is making great strides in creating more comprehensive language models that better serve the black diaspora, but current AI models, from ChatGPT to Gemini, cannot fully support the more than 2,000 languages spoken across Africa.
Yinka Iyinolakan created CDIAL.AI to solve this problem. CDIAL.AI is a chatbot that can speak and understand almost all African languages and dialects, with a particular focus on speech patterns rather than text.
Iyinolakan echoed the same sentiments to TechCrunch as many black Americans. This means that the underlying AI models are primarily scraped from internet data and the most commonly used languages. Many African languages and traditions, like those of African American descent, do not exist on the Internet because they are cultures that were historically transmitted orally rather than in written form. This means that AI models do not have enough information about African culture to train themselves, creating a knowledge gap.
For CDIAL.AI, Iyinolakan brought together more than 1,200 native speakers and linguists from across Africa to gather knowledge and insights to build “the world’s first multilingual speech-first large-scale language model.” The company plans to expand to include more languages in the next 12 months and build a model that supports text, voice, and images.
He is not alone here. Google recently provided Kenya-based Jacaranda Health with a $1.4 million grant to build machine learning services to work with more African languages, and Intron Health recently announced that it would provide research on more than 200 accents spoken across Africa. We've raised millions of dollars to expand clinical speech recognition. .
“Silicon Valley wants to believe that it is the be all and end all of artificial intelligence,” Iyinolakan said. “But for every company to ‘get’ artificial intelligence as its North Star, it must involve one-third of the world’s knowledge.”
make progress
Introducing AI chatbots isn’t the only innovation that Black founders are trying to solve.
Steve Jones started a company called pocstock to create stock images of people of color. That's because there has been a lack of minority representation in stock images for decades. This is one of the reasons why today's models spit out predominantly white images when users ask them to create pictures of everything from doctors to pop singers.
“All platforms and tools must be trained with complete, racially inclusive, and culturally accurate data,” Jones told TechCrunch. “Otherwise we will (perpetuate) the bias issues that the larger society currently faces.” To solve this problem, pocstock has spent the last five years creating its own visual tagging system that collects diversity data and contributes to a database that helps companies train AI models to create more comprehensive imaging. .
However, some improvements are being made. Jones said he's found that larger stock imaging companies that source AI companies are making more progress in increasing the diversity of their content. Pasmore also sees a brighter future ahead, saying that personalized AI is the future anyway, and that the more AI models interact with users, the better they will understand the needs and requirements of specific individuals. “I think it eliminates a lot of prejudice.”
There may be room for more culturally specific AI models in the future, especially as more Black-owned alternatives continue to emerge. After all, the world is wider and more nuanced. There is no purpose to contain it in one black box.
“We hope to see more entrepreneurs of color develop their own AI platforms or participate in creating new AI-related jobs in the early stages of the next economic boom,” Jones said. “AI will create trillionaires. “I want to see people of color establishing themselves as producers rather than just consumers.”
This document has been updated to reflect training from Spark Plug.
