AI

AI is the next frontier — but for whom?

Comment

3D rendered classic sculpture Metaverse avatar with network of low-poly glowing purple lines. Machine learning and artificial intelligence concept. Animated 3D NFT artwork example. Web 3.0 technology background.
Image Credits: salihkilic / Getty Images

A few weeks ago, a founder told me it took three hours of endless clicking to find an AI-generated portrait of a Black woman. It reminded me, in some ways, of a speech I saw three years ago when Yasmin Green, the then-director of research and development for Jigsaw, spoke about how human bias seeps into the programming of AI. Her talk and this founder, miles away and years apart, are two pieces of the same puzzle.

Discussions about diversity are more important than ever as AI enters a new golden era. Every new technology that appears seems to be accompanied by some harrowing consequence. So far, AI has contributed to racist job recruiting tactics and slower home approval rates for Black people. Self-driving cars have trouble detecting dark skin, making Black people more likely to be hit by them; in one instance, robots identified Black men as being criminals 9% more than they did white men, which would be put under a new light if judicial systems ever begin adopting AI.

AI ethics is often a separate conversation from AI building, but they should be one and the same. Bias is dangerous, especially as AI continues to spread into everyday life. For centuries, doctors once judged Black people on criteria now deemed racist, with one prevalent belief being that such people experienced less pain. Today, algorithms discriminate against Black people; one study from 2019 found that an algorithm used by U.S. hospitals “was less likely to refer Black people than white people who were equally sick to programs that aim to improve care for patients with complex medical needs.”

Right now, bias appears in various AI subsectors, ranging from investment to hiring to data and product execution, and each instance of bias props up others. Eghosa Omoigui, the founder of EchoVC Partners, told TechCrunch that though AI can be “incredibly powerful,” society is still far from “flawless” artificial intelligence.

“This means that the likelihood of AI bias in outcomes remains high because of the excessive dependencies on the sources, weights and biases of training data,” he said. “Diverse teams will prioritize the exquisite understanding and sensitivity necessary to deliver global impact.”

Omoigui’s brother, Nosa, the founder of ESG compliance regulator Weave.AI, reiterated that point. Many of these models are black boxes, and creators have no particular insights into the inner workings of how a prediction or recommendation is achieved, he said. Compared to Wall Street, AI is practically unregulated, and as the level of governance fails to match its growth, it risks going rogue. The EU proposed steps to reduce and account for bias in AI-powered products, with some pushback, though the proposition itself puts it slightly ahead of where the U.S. is now.

In fact, Eghosa said many investors don’t care or think about diversity at all within AI and that there is a groupthink mentality when it comes to machine-led capabilities. He recalled the reactions investors gave him when he helped lead an investment round for the software company KOSA AI, which monitors AI for bias and risks.

“Quite a few investors that we spoke to about the opportunity felt very strongly that AI bias wasn’t a thing or that a ‘woke product’ wouldn’t have product-market fit, which is surprising, to say the least,” Eghosa said.

It should be no surprise that investors don’t think bias is a problem, given that they show it themselves in their investment patterns, particularly in AI. Crunchbase data shows that in 2019, U.S.-based companies with at least one Black co-founder raised $40 million out of the total $19.95 billion in venture funds allocated to U.S.-based AI companies. U.S.-based companies with at least one Black co-founder picked up $308 million out of $41.57 billion in 2021 and $54 million out of $23.48 billion in 2022.

Data visualization by Miranda Halpern, created with Flourish

The capital amount allocated for companies with at least one Black co-founder has hovered under 1% for the past few years, with no clear indication that it is increasing anytime soon, compared to companies with at least one woman founder, for example, which have seen a year-over-year funding percentage increase since 2019.

The lack of investment in new companies with Black founders is aided by the lack of diversity in AI teams at existing companies, which, in turn, impacts product execution. All of this, combined with the lack of governance, creates a self-perpetuating cycle. Parfait, an AI-powered wig company, targets the expensive and time-consuming needs of managing Black hair. Its co-founder, Isoken Igbinedion, said investors were intrigued by her product, especially since using AI within that industry wasn’t common.

Calling AI a numbers game, Igbinedion said that if there isn’t enough robust representation of all types of people in datasets, then resulting applications won’t work well for everyone.

“This is especially frightening for Black women who are typically the most underrepresented,” she told TechCrunch. As a Black founder, she was able to create a product that could cater to her community as she understood its needs; such nuance is not something often seen or prioritized by the white men building these robust AI companies.

Igbinedion said solving the problem of AI bias requires more than the ability to bootstrap; there needs to be real support that comes with real dollars that can invest in long-term solutions. Once again, Black founders are willing to fill the gaps, but in this industry, there is not enough attention being paid as to why it is always diverse individuals who must stand up and ignite change.

Parfait is spending time working to construct datasets to help others build equitable training models — basically writing better recipes to feed machines. Nosa implemented a knowledge graph operated by an algorithm with human oversight to prevent his system from making any “random” suggestions.

“As AI pervades society, it is critical for developers and corporations to be good stewards of this technology but also hold fellow technologists accountable for these unethical use cases,” Igbinedion added.

While chatting with TechCrunch, Ekechi Nwokah, the founder of the embedded lending platform Migo, raised an interesting point. He said there is so much excitement in this space right now — people are building so fast and investors are so bullish — that there often isn’t any time to put in the work to look for a diverse team or to parse through algorithms to make everything perfectly unbiased or even just less biased than it is now. Especially in the early stages, entrepreneurs want to get their products out, he said, adding, though, that as a business matures, there is an increased likelihood that founders will backtrack to fix any problems.

“I think most folks that are working on AI models have good intent,” Nwokah told TechCrunch. He said he tries to prevent bias by focusing on hiring and implementing policies that have diversity as a core pillar of his company, hoping it then leads to better decision-making regarding how products are built.

At the same time, it takes time to fine-tune models and unweave the historical cause-and-effect input data given to these AI models, Nwokah continued. “AI models predict the future. It’s a very difficult thing to do, and you can only predict the future based on what’s happened in the past, so the only way the model is gonna predict is based on what it’s seen before, just like a human being,” Nwokah said.

Perhaps then, to combat the biased narratives fed to some AI models, the builders must stitch a representation of a new society, an ideal one. As Nwokah insinuated, the past has been written, and the builders of AI must envision and implement a better future.

Eghosa, meanwhile, said he worried that this generative AI “frenzy” is causing people to forget the requisite respect, recognition and guardrails that AI-infused tools and products should bear.

As in other areas of entrepreneurship with diverse founders, he said underrepresented founders already prioritize equitable building and execution. For some reason, other founders cannot master this multitasking, which is why it took three hours to find an AI-generated portrait of a Black woman.

The concept of the generative model was great, but the algorithm was trained from a mainstream photo-sharing site and inherited its biases. The photos on that popular site had many images of white women, meaning it was easier for the algorithm to create believable white faces than for other races. That system is a few years old, but the problems that produced it still plague the industry.

“I don’t know if we will ever completely get rid of bias,” Eghosa said. “That said, I am excited that there are — and I am confident more are on the way — potential scalable approaches powered by AI to identify it, analyze it and offer data-driven decision-support tools and platforms to mitigate or eliminate it. Its power and utility are widely underestimated. With requisite guardrails, it will empower and turbocharge decision-making, productivity and economic growth while creating more access to opportunities and a societal lift.”

More TechCrunch

Looking Glass makes trippy-looking mixed-reality screens that make things look 3D without the need of special glasses. Today, it launches a pair of new displays, including a 16-inch mode that…

Looking Glass launches new 3D displays

Replacing Sutskever is Jakub Pachocki, OpenAI’s director of research.

Ilya Sutskever, OpenAI co-founder and longtime chief scientist, departs

Intuitive Machines made history when it became the first private company to land a spacecraft on the moon, so it makes sense to adapt that tech for Mars.

Intuitive Machines wants to help NASA return samples from Mars

As Google revamps itself for the AI era, offering AI overviews within its search results, the company is introducing a new way to filter for just text-based links. With the…

Google adds ‘Web’ search filter for showing old-school text links as AI rolls out

Blue Origin’s New Shepard rocket will take a crew to suborbital space for the first time in nearly two years later this month, the company announced on Tuesday.  The NS-25…

Blue Origin to resume crewed New Shepard launches on May 19

This will enable developers to use the on-device model to power their own AI features.

Google is building its Gemini Nano AI model into Chrome on the desktop

It ran 110 minutes, but Google managed to reference AI a whopping 121 times during Google I/O 2024 (by its own count). CEO Sundar Pichai referenced the figure to wrap…

Google mentioned ‘AI’ 120+ times during its I/O keynote

Firebase Genkit is an open source framework that enables developers to quickly build AI into new and existing applications.

Google launches Firebase Genkit, a new open source framework for building AI-powered apps

In the coming months, Google says it will open up the Gemini Nano model to more developers.

Patreon and Grammarly are already experimenting with Gemini Nano, says Google

As part of the update, Reddit also launched a dedicated AMA tab within the web post composer.

Reddit introduces new tools for ‘Ask Me Anything,’ its Q&A feature

Here are quick hits of the biggest news from the keynote as they are announced.

Google I/O 2024: Here’s everything Google just announced

LearnLM is already powering features across Google products, including in YouTube, Google’s Gemini apps, Google Search and Google Classroom.

LearnLM is Google’s new family of AI models for education

The official launch comes almost a year after YouTube began experimenting with AI-generated quizzes on its mobile app. 

Google is bringing AI-generated quizzes to academic videos on YouTube

Around 550 employees across autonomous vehicle company Motional have been laid off, according to information taken from WARN notice filings and sources at the company.  Earlier this week, TechCrunch reported…

Motional cut about 550 employees, around 40%, in recent restructuring, sources say

The keynote kicks off at 10 a.m. PT on Tuesday and will offer glimpses into the latest versions of Android, Wear OS and Android TV.

Google I/O 2024: Watch all of the AI, Android reveals

Google Play has a new discovery feature for apps, new ways to acquire users, updates to Play Points, and other enhancements to developer-facing tools.

Google Play preps a new full-screen app discovery feature and adds more developer tools

Soon, Android users will be able to drag and drop AI-generated images directly into their Gmail, Google Messages and other apps.

Gemini on Android becomes more capable and works with Gmail, Messages, YouTube and more

Veo can capture different visual and cinematic styles, including shots of landscapes and timelapses, and make edits and adjustments to already-generated footage.

Google Veo, a serious swing at AI-generated video, debuts at Google I/O 2024

In addition to the body of the emails themselves, the feature will also be able to analyze attachments, like PDFs.

Gemini comes to Gmail to summarize, draft emails, and more

The summaries are created based on Gemini’s analysis of insights from Google Maps’ community of more than 300 million contributors.

Google is bringing Gemini capabilities to Google Maps Platform

Google says that over 100,000 developers already tried the service.

Project IDX, Google’s next-gen IDE, is now in open beta

The system effectively listens for “conversation patterns commonly associated with scams” in-real time. 

Google will use Gemini to detect scams during calls

The standard Gemma models were only available in 2 billion and 7 billion parameter versions, making this quite a step up.

Google announces Gemma 2, a 27B-parameter version of its open model, launching in June

This is a great example of a company using generative AI to open its software to more users.

Google TalkBack will use Gemini to describe images for blind people

Google’s Circle to Search feature will now be able to solve more complex problems across psychics and math word problems. 

Circle to Search is now a better homework helper

People can now search using a video they upload combined with a text query to get an AI overview of the answers they need.

Google experiments with using video to search, thanks to Gemini AI

A search results page based on generative AI as its ranking mechanism will have wide-reaching consequences for online publishers.

Google will soon start using GenAI to organize some search results pages

Google has built a custom Gemini model for search to combine real-time information, Google’s ranking, long context and multimodal features.

Google is adding more AI to its search results

At its Google I/O developer conference, Google on Tuesday announced the next generation of its Tensor Processing Units (TPU) AI chips.

Google’s next-gen TPUs promise a 4.7x performance boost

Google is upgrading Gemini, its AI-powered chatbot, with features aimed at making the experience more ambient and contextually useful.

Google’s Gemini updates: How Project Astra is powering some of I/O’s big reveals