How firms like LivePerson, Grammarly, Adobe, Darktrace, and others are using AI to enhance people's lives in novel ways.
The 7 most creative artificial intelligence firms in 2022
Explore the whole list of Fast Company's Most Innovative Companies for 2022, which includes 528 companies that are redefining their sectors, businesses, and culture.
We've chosen the organizations making the most of their endeavors across 52 categories, including the most creative AR/VR, design, and security enterprises.
Artificial intelligence, once a fancy buzzword, is fast evolving into something far more valuable: an everyday reality that quietly enhances a wide range of experiences.
Our list of the most innovative AI businesses for 2022 reflects this. Many of them have been working on it for a long time, and their efforts reflect both development and revolution.
Customer care platform LivePerson, for example, debuted its first AI-powered bots in 2018. Its "conversational AI" now handles millions of consumer interactions for enterprises selling everything from doughnuts to diamonds after many versions.
Grammarly is another option. Since 2009, it has used machine learning to interpret textual messages, detect problems, and offer recommendations.
But it's only recently that technology has advanced sufficiently to allow Grammarly to detect truly minor readability and style errors, and then suggest complex adjustments to remedy them.
In our list, such fine-tuned AI applications coexist with bold new initiatives.
Eightfold AI's Career Hub, for example, analyzes data to assist employees in locating mentors and other career-enhancing tools within a firm.
Immunai is combining math and biology to map the human immune system, a potentially game-changing step toward more efficient medication development.
With AI's rising capacity to do valuable tasks, it also has the potential to cause harm, whether through intentional exploitation or due to biased or otherwise faulty data. That's why firms like Adobe are recognizing and confronting this unsettling truth.
Before releasing new features like Photoshop's new neural filters, the company conducts an ethics assessment to identify—and eliminate—any problematic components.
LivePerson
Bots are being taught to assist human agents.
Conversational AI from LivePerson allows businesses to automate routine customer care activities via online chat and text messaging, freeing up professional agents to focus on more complex issues.
In 2021, the firm expanded its original focus on after-purchase service by introducing the capacity to incorporate Conversational AI into commerce platforms.
For example, in 9,000 stores, Dunkin' has put QR codes on food packaging, allowing consumers to sign up for its loyalty program by conversing with a bot.
Conversational AI's new region isn't limited to commerce: Bella Health, a COVID-19 screening bot in use at 500 locations, assists in the detection of illnesses before they spread to coworkers unintentionally.
A new feature called AI Annotator has allowed support staff to enhance a company's bots on the fly, without requiring a profound understanding of data science; altogether, automated interactions done on LivePerson's platform have increased by 40% year over year.
OPENAI
OpenAI unveiled GPT-3, a groundbreaking AI tool that can write humanlike language and computer code, in May 2020. However, it did not make the technology widely available until November 2021, after it had put in place controls to prevent it from being used in potentially harmful ways.
It began allowing third-party developers to adapt GPT-3's algorithms for their own purposes a month later.
In addition, in 2021, the business unveiled three key new technologies that expand on GPT-3's achievements. Codex converts plain language into computer code, potentially democratizing software development.
Dall-e transforms textual descriptions into visuals ("an avocado-shaped couch").
Clip may also detect the general theme of a collection of photographs (for example, "country line dance") without being trained to recognize specific subjects.
GRAMMARLY
For giving ever-more-insightful writing guidance.
Thanks to powerful AI, Grammarly has moved well beyond assisting individuals in avoiding fundamental grammatical errors.
The writing tool can now check for tone, fluency, and general readability of content. It now recommends full-sentence rewrites that keep the original meaning, rather than just highlighting flaws.
Grammarly has also embraced nonbinary pronouns and neopronouns to keep up with changes in the way people use language and has begun checking for culturally stigmatizing words like "Chinese virus."
In 2021, the firm introduced additional capabilities for its 30,000 office users, such as brand tones and style guidelines, that allow them to apply various priorities depending on the context of a document.
In September, Grammarly for Developers was released, allowing software developers to incorporate Grammarly suggestions into any online project with just a few lines of code.
LinkSquares
LinkSquares employs machine learning to automate the process of contract interpretation, removing the need to keep track of everything from terms and conditions to legal duties and renewal dates manually.
Customers may save time and money by automating risk assessments, privacy audits, and other activities that require a complete grasp of the contractual landscape—all without having to hire expensive outside counsel.
Due to new connections with Salesforce and DocuSign, LinkSquares' clients were able to incorporate its solutions into their everyday processes in 2021.
New capabilities related to crucial challenges, such as maintaining compliance with the California Consumer Privacy Act, have also been added to the company's AI.
After growing by more than 1,000 percent in just two years, LinkSquares has analyzed over 3 million contracts and collected over 40 million data points.
ADOBE
for making Photoshop magic more accessible.
Adobe's new neural filters leverage artificial intelligence to make visual effects that previously required hours of effort and years of image-editing ability as simple as point-and-click.
By using them, you can easily transform the expression of a photo subject from deadpan to happy.
Alternatively, change the direction in which someone is gazing. Alternatively, subtly colorize a black-and-white photograph.
The filters, which are part of Adobe's "Sensei" AI portfolio, employ generative adversarial networks, an advanced kind of machine learning.
As a result, they may do things like depict areas of a face that weren't previously visible when editing a portrait.
The neural filters, like all new Sensei capabilities, were authorized by an Adobe ethics committee and review board, which evaluates AI products for concerns such as skewed data.
This procedure uncovered a problem with how particular hairstyles were represented in the instances of these filters and repaired it before they were distributed to the public.
CITRINE INFORMATICS
For the purpose of accelerating the creation of superior products.
Citrine aids chemical and material businesses in reducing the time and risk required in developing new goods that are more sustainable and less hazardous.
When compared to traditional approaches based on trial and error, the company claims that their platform can accelerate R&D efforts by up to 98 percent.
It incorporated AI-powered capabilities in 2021 to estimate the performance of new components and determine if different research pathways will meet key product development objectives.
A polymer manufacturer developed a new, lighter material for automotive applications, while a specialized chemical firm removed a hazardous element from its product portfolio, resulting in a 500% increase in weekly active users year over year.
SCALE AI
Better data is necessary for getting AI properly.
When the methods that are used to train machine learning models are presented with data that isn't normal, they frequently fail. Scale Nucleus,
a new product from Scale AI, is meant to assist businesses in picking the optimal training data in the first place, rather than scrambling after the fact to deal with concerns like inadvertent gender bias in findings.
Scale Nucleus has been utilized by over a hundred companies to label over 25 million pieces of data.
According to Scale AI, Toyota saw a tenfold boost in annotation throughput in only a few weeks.
The company's technology has a wide range of applications, as seen by its customer list, which includes PayPal, Pinterest, and Etsy, as well as the United States Air Force.