When the first wave of modern-day AI hype started, a lot of people I knew were worried. Much of that worry has, so far and unfortunately, been well-founded. AI has done a lot of what people were afraid it would do. It’s pushed people out of jobs, consumed massive amounts of water and energy, and warped our ideas of reality, friendship, and community. Bad look, all around.
I can’t remember when I personally became aware of AI’s rapid spread—probably 2023, like most nontechnical people. What I do remember are the key moments since then when AI seemed to have an outsized impact on my life: getting laid off; being subsequently told by my mentor I should consider career paths outside of writing; struggling to find work when, as my mentor predicted, much of the job market for creatives was rocked by the generative AI boom; late-night despair-ridden, anxiety-laden, Wing Stop-fueled conversations with friends about our futures, society’s future, and the earth’s future; and now, when much of my writing for Stack Overflow has—unsurprisingly—been about AI.
A few years ago, I learned about a woman who was developing an AI tool to help people appeal insurance claims, borne from her own experience of being denied for small, sometimes incomprehensible reasons. Her theory was that an AI tool—trained on data about successful appeals—could help the everyday consumer take on the mentally taxing and tedious task of appealing insurance denials.
This project—ultimately a labor of love for creator Holden Karau and her business partner Melanie Warrick—is now the completely free tool FightHealthInsurance. To this day, it’s helping people appeal bogus insurance denials.
It was the first time I saw someone using AI just to help people. Not to make a quick buck, or increase productivity, or start a viral sensation—just to help people. As the meme goes, “Great use of free will.” In a time when the only thing my friends and I seemed to talk about was how bad AI would become if someone didn’t stop it (and how powerless we were to stop it, just a bunch of 20-somethings barely scraping by), here was a person who could adapt, take what was being offered by the world, and turn it around to do something actually good, just because they could.
From that point on, I would use FightHealthInsurance as a rebuttal to the doom-and-gloom approach to AI my friends would take. I’m a pretty clear-cut optimist, which can sometimes get on my loved ones’ nerves. FHI was a strong weapon in my favor—it’s hard to counter the argument that it’s helping people, at scale, in a way that wouldn’t be possible without AI. My friends’ rebuttals would usually be along the lines of, “Well one good thing doesn’t negate all these other bad things that AI is causing.” Which is true.
But what about a bunch of good things? What if we all did enough good things to make a dent in the bad?
The good news: Holden’s good work with AI is not the only example out there, and I’m here to tell you about a few of them happening, here and now. Not all the organizations I’ll be featuring have the same highly personal and selfless backstory as FightHealthInsurance, they all share one thing in common: they’re taking our modern technological revolution in stride and using it for the good of humanity.
…And what exactly does it mean for AI to do good?
Ah, such small and uncomplicated questions I write about. For me, in order to be considered for the “AI for good” category, an organization must deeply embed AI technology into their work, either as a major function or as a way to scale. Do-goodery is a bit harder to define, but I’m calling it work whose mission is solely focused on solving humanitarian issues, preserving the environment/fighting climate change, or helping the everyman. These things can range from detecting earthquakes for better disaster response to reducing textile waste with AI inspection to assisting everyday people with insurance claim appeals. Ultimately, the work needs to positively impact people, the earth, or our society to be considered as “doing good.”
Even with that fairly specific scope of “doing good,” the possibilities that AI affords do-gooders is limited only by their imaginations. As Ryan Panchadsaram, Technical Adviser at Kleiner Perkins, succinctly told me, “My perspective on AI is that it is an incredible technology that can unlock so many areas of good for society.”
To illustrate this, let me tell you about Canary Speech. While AI-augmented healthcare is nothing new—and clearly on its way to becoming an integral part of our medical system, if OpenAI’s newest product is to be believed—what Canary Speech does is different. Canary Speech was founded ten years ago, when former NIH researchers Jeff Adams and Henry O’Connell—who would both go on to have illustrious commercial careers doing things like working on Siri, leading HP, and inventing Alexa after leaving the NIH—decided to do just one more thing with their expertise of voice: use it to diagnosis disease. Today, their AI tech is able to identify vocal biomarkers in patients to help clinicians detect and diagnose disorders and diseases like Alzheimer’s, Parkinson’s, depression, and anxiety.
This kind of work isn’t just supported by AI—it only exists because of AI. From far before the current AI boom, Adams and O’Connell were using neural network machine learning to research and identify these vocal biomarkers. They’ve now identified over 2,500 of these speech features. And if you’re wondering just how much of this is real, you can ask their recently published peer-reviewed study with the University of Massachusetts Chan Medical School on the effectiveness of AI voice-screening for Parkinson’s detection.
What’s most meaningful to me is the human experiences that the team at Canary Speech shared. During a routine postpartum check-up, a new mother told her clinician that she “felt fine,” but her Canary Speech score came back with high levels of depression and anxiety. Based on her score, the clinician took the time to discuss her wellbeing with her on a deeper level, and this conversation allowed the new mother to open up and share that she had been struggling, quite a lot, since the birth. She was able to get care for her postpartum depression, something that might not have happened without Canary Speech’s AI.
Here is the good that is possible when we give AI tools to human experts, who wield them for the benefit of others. Canary Speech is, under FDA guidelines, a clinical decision support tool. It gives medical professionals a new way to identify patient needs that are sitting under the surface. In the case of the new mother, Canary Speech’s AI prompted the human clinician to have a human conversation with the patient, ultimately leading to medical care. As the saying goes: a canary in the coal mine. An early warning system, only possible because Canary Speech put powerful AI in the right hands.
It’s easy to see how this AI tool could benefit overloaded hospitals, underserved healthcare deserts, and patients without access to affordable healthcare. In just one visit, vulnerable populations (like elderly people with Parkinson’s) or those who may be reluctant to access care for themselves (like busy, overwhelmed new parents) can get potentially life-saving diagnoses, reducing the time and energy needed to receive necessary healthcare. And this early detection system has only been made possible because of AI.
Canary Speech isn’t the only company combining AI and human expertise in the name of good. Ryan Panchadsaram, who is one of the leaders of Speed & Scale’s climate action plan, has seen experts across industries “apply [AI] to all of society’s most pressing problems: climate, healthcare, education.”
This sort of expertise is what drives much of the work of the fellows from AWS’ Compute for Climate. Lisbeth Kaufman, Head of Climate Tech Startups BD at AWS, told me, “The founders in the Compute for Climate Fellowship show that when AI is designed thoughtfully with ethics, efficiency, and impact at the center, it can be a powerful force for humanitarian and planetary good.” I was lucky enough to get to connect with two of the founders in their fellowship, Aigen and Smartex, who are both working to lessen the impact humans are having on our earth and climate.
Aigen’s expertise—and that of its founder and CEO Kenny Lee—is in the agritech space, helping small, family farms automate their processes and ultimately stay alive using autonomous agriculture robots. On small farms, mistakes are costly. Lee told me, “[The] average age of the farmer in the US is close to 60 years old, holding on to operations their families built over generations. They are one bad season away from losing everything.” According to Lee, “[Small farms] can’t afford to hire enough labor. They can’t compete with industrial-scale operations. And there’s no one to pass the farm to because the economics don’t work for the next generation.”
That’s where Aigen comes in. According to Lee, Aigen’s agriculture robots are able to operate at scale on both large and small farms, “running 12+ hours daily on solar power alone, making millions of real-time decisions per hour in environments that change constantly.” These real-time decisions are made possible through Aigen’s physical AI that works on the “edge,” meaning the AI can process everything directly in the hardware, circumventing the need for cloud computing and increasing the speed of analysis and decision-making.
From my own perspective, one of the most exciting use cases of Aigen’s tech is their computer vision for weed control. Because their physical AI is able to recognize and catalogue plants with speed and accuracy difficult for humans, their robots are able to mechanically weed vast fields, replacing the need for chemical sprays, which harm wildlife and native plants. And it’s only possible because of AI. “Farms are messy, unpredictable environments. Without AI, we’d be stuck blanket-spraying chemicals or hand-weeding at costs farmers can’t afford…AI is the only reason this works,” Lee said.
Aigen’s tech, which includes plant recognition through thousands of plant images per minute, gives these farmers a fighting chance in an agricultural economy leaning further and further into monopolization by a few large corporations. Ultimately, Aigen’s work aims to help democratize access to advanced farming tools, taking on the backbreaking, costly, and dangerous work that farmers must do to upkeep their fields—like weeding in 110-degree heat. “The family farm in North Dakota should have the same access and capabilities as the industrial operation in California,” Lee said.
Aigen isn’t the only one making a real impact on the environment. Smartex, another fellow from AWS’ Compute for Climate, is using AI to reduce textile waste, something that has been a growing environmental issue since the rise of fast fashion.
Max Easton, Smartex’s CEO, shared that their work is fundamentally changing the way we manufacture textiles. They integrate their technology directly into the factory machinery to reduce waste that was at one time an unavoidable part of mass garment production. Because much of the textile waste that comes out of factories is because of errors caught too late in the manufacturing process—for instance, defects that aren’t caught until dyeing and cutting has been done to the garments—Smartex’s early defect detection system allows factories to save resources and reduce waste. “We inspect 100% of fabric production in real time, automatically detect defects, stop repetitive issues, and generate objective quality data for every roll produced,” Easton explained.
This process turns quality control from a “reactive checkpoint into a preventive system,” Easton said, which uses less energy, consumes less water, and lowers CO2 emissions for factories. This is not just one or two factories, by the way—they’re able to deploy this technology globally, and every time a single model improves, so do the rest of them, something that Easton calls a “multiplier effect” which translates “into thousands of tonnes of fabric waste avoided.”
For Easton, “AI has enormous potential when it is applied to invisible, systemic problems, especially in industries that employ millions of people but have been historically under-invested in technology.” Often that means helping underserved communities that are doing hard, laborious, and dangerous work, and getting underpaid doing it. “In manufacturing,” Easton says, “AI can improve working conditions by removing guesswork and firefighting from daily operations. When factories become more efficient and predictable, they are more resilient, economically and socially.”
Like Aigen, Smartex’s work is only possible because of their AI. Easton shared, “AI is inseparable from both our product and our mission. Without AI, real-time inspection at full production speed simply wouldn’t be possible. More importantly, the data foundation we create is objective, real-time, and covers 100% of output, which only exists because AI captures and structures it automatically.”
Aigen and Smartex are moving the needle in the fight against climate change, along with other Compute for Climate fellows like Realta Fusion. It’s a fight we can’t afford to lose, but one that we are losing. Compute for Climate’s Kaufman told me, “The climate crisis is moving faster than traditional systems can keep up. We’re seeing extreme weather, food and water stress, and energy challenges collide. At the same time, many of the most promising solutions just aren’t possible without serious computing power.”
For the Compute for Climate fellows, AI has improved the speed, efficiency, and precision of their work, pushing these innovative technologies forward at a pace closer to what is necessary to save the planet. Kaufman told me, “Many of these startups are tackling inherently global problems, so AI allows them to move beyond pilots and operate at a much larger, even planetary level.”
This is evident in both the cases of Smartex and Aigen, whose technology relies on the constant learning and fine tuning of models deployed across numerous farms and factories. “We see it dramatically speed up research and development, whether that’s running fusion energy simulations, improving extreme weather forecasting, or accelerating advances in agriculture. What used to take years can now happen in weeks,” Kaufman said.
Microsoft’s AI for Good Lab is also supporting the efforts of researchers and NGOs working to advance sustainability, whether it be through addressing the climate crisis or bolstering conservation efforts. I visited their activation at Microsoft Ignite last year, where I learned about their work to save the Amazon rainforest in partnership with the Alexander von Humboldt Institute. By using the power of AI, they’ve been able to fight illegal deforestation, monitor endangered and protected wildlife, and identify species through AI bioacoustics, all key elements for preserving the biodiversity of the Amazon rainforest.
In the same way that AWS’ Compute for Climate program addresses both climate change and humanitarian issues through their work, Microsoft’s AI for Good Lab supports research aimed at protecting the fundamental rights of human beings—something that may seem a bit paradoxical to those who believe AI is the downfall of humanity. Much of the important work of protecting basic human rights—things like access to food, clean water, and livable conditions—comes in the form of early detection systems, something that AI has supercharged. In partnership with organizations like Planet, Catholic Relief Services, and SEEDS India, the AI for Good Lab has developed detection systems for things like disaster resilience, hunger forecasting, and extreme-heat mapping, often focused on at-risk communities and lower-resourced countries. The benefits of this technology are palpable: it resulted in faster disaster-relief in post-earthquake Afghanistan, proactive intervention during a period of food insecurity in southern Malawi, and precise Red Cross responses during the Lahaina fires on Maui.
But you don’t need the backing of AWS or Microsoft to do good with AI. All of us can do it. We just need to put these powerful tools in the right hands—our own. In the next part of this series, some of my favorite AI do-gooders share how the everyday you and I can use AI for good.
