Artificial intelligence has grabbed headlines for the past few years, but too often the press oversells the risks and rewards of AI. We read about AI’s inevitable bias, and its deadly use in war. Of course, we also read the positive, like a Google computer beating the world’s best Go players.
But these stories fail to accurately reflect the best uses of AI today. I wrote years ago that IBM needed to stop pitching its Watson as a miracle cure for most everything, and instead position it for more pedestrian tasks. In like manner, we’d do better to celebrate AI adoption in small steps that add up to major savings–like food and waste and other sectors.
How some of these emerging AI vendors got here with solutions that actually work-as-advertised offers important lessons to any business. They offer classic examples of real disruption in formerly staid markets…one incremental step at a time.
SEE: Artificial intelligence ethics policy (TechRepublic Premium)
Getting real with AI
The term AI is so broad it is almost useless to an average business, so it’s better to be specific. Where AI is quickly gaining traction in “normal” industries is better called supervised reinforcement learning. Feed enough annotated data (you, a service or special software tags the data with a description) into a machine learning (ML) algorithm (usually free and open source) and run the results through your application (usually on a public cloud), and suddenly your organization is solving hard problems faster with more accuracy. Because the ML “learns” as it goes, you create an iterative loop to feed more new data as well as corrected old data adjusted for new information, bias or other problems back into the loop.
It keeps getting better.
In the end, it’s the data that powers AI models. The more they gather, like the user-generated content that powers the world’s most popular social platforms, the more useful and powerful their AI engines become. Global tech giants like Facebook, Google, Tesla and more can and have created competitive moats through their increasing leads in AI.
But everyday companies can take advantage of AI, too. The more than six million Java developers today can now run their programs through an AI-powered software testing system from Diffblue based on the same AI concepts Google used to beat the best Go players. It automatically creates Java unit tests 100X faster than humans with 85% code coverage. Developers can save 20% to 50% of their time previously sucked up by tedious test-writing tasks.
How about mundane tasks for any company that sells something (which is to say, pretty much every company)? Scoring sales leads is typically an arcane art, with expensive, complicated systems to learn, or a reliance on the intuition of a sales team. AI startup Akkio can suck up your Excel spreadsheets and spit out cutting-edge, high-quality predictive analytics dashboards in a day or two. To achieve the same results would normally require a dedicated app team and an AI scientist, and would likely take months.
Or consider education. Startup Riiid uses AI to improve education, and created a smartphone app that was free for the first two years but good enough that it collected millions of data bytes from users who wanted to improve their scores on a standardized test of English proficiency. The app could predict with better than 95% accuracy your score after just 10 minutes but then also created a personalized learning program to boost your final scores by 10% to 20% on average. Riiid also released the biggest public dataset on education and used that data to sponsor the largest Kaggle competition in 2020 to attract the world’s smartest AI researchers to improve the algorithm. Riiid is now accelerating the rollout of its AI engine to K-12 schools around the world, as well as to corporate employee development programs, after announcing a $179 million funding round.
Unsung AI heroes
You probably haven’t heard of any of these companies, right? Good AI projects start small and then build. Success in AI is modest, incremental. It also helps to avoid hiring expensive dedicated teams of AI and data scientists. Companies like Scale AI, which offers a services-led platform, and Labelbox, which offers a software-led platform, can show companies how to be successful with AI and not fumble in the dark making costly mistakes that disappoint the C-suite.
I’ll finish with food and waste, two boring problems that are terribly important for the future survival of our species and the health of our planet.
Research consultancy McKinsey estimated in 2019 that reducing food waste through AI could be as much as a $127 billion market opportunity by 2030. In a wonderful and recent report written for kids, McKinsey estimated that if all the world’s food production represented a 30-slice loaf of bread, we collectively waste 12 slices a day. To keep up, food production will have to nearly double as the global population reaches 10 billion by 2050, even as land under cultivation shrinks.
Can AI really help? It already is. Here are some current examples of Labelbox customers–companies need to annotate mountains of data to train their AI systems–putting AI to work to put food on our tables and cut waste.
Ireland-based Cainthus uses computer vision to monitor livestock herds 24/7 and send alerts to farmers to feed more cattle with less.
Silicon Valley-based Everest Labs uses AI and robotics to reduce waste and develop environmentally-friendly products. Its robot can sort 60 cartons of recyclables per minute.
John Deere’s Blue River Technology unit sells smart tractors that can spray herbicides precisely, cutting costs, increasing yields and reducing pollutants in the environment.
U.K.-based Winnow Solutions uses computer vision and AI to track and analyze food waste in industrial kitchens. Its customer IKEA slashed food waste 45% in just three months, and the company estimates its solution has reduced customer CO2 emissions by more than 60,000 tons to date.
Netherlands-based Xarvio Digital Farming Solutions uses smartphone, drone and satellite imagery to build AI-powered products that advise farmers on how to maximize productivity.
This is how AI will take over the world, through incremental steps that add up to make food production, education and many other industries more efficient and effective.
Disclosure: I work for AWS, but the views expressed herein are mine.
Editor’s note: Details about Riiid’s rollout of its AI engine were corrected.