Enterprise

OpenAI rival AI21 Labs raises $64M to ramp up its AI-powered language services

Comment

Binary background with locks representing security or encryption.
Image Credits: JuSun / Getty Images

The enterprise is bullish on AI systems that can understand and generate text, known as language models. According to a survey by John Snow Labs, 60% of tech leaders’ budgets for AI language technologies increased by at least 10% in 2020. And one vendor, OpenAI, says that its premiere language model, GPT-3, is being used by tens of thousands of developers.

Eager for a slice of the pie, new providers have materialized in recent years claiming to bring unique language modeling capabilities to the table. Beyond well-resourced startups like OpenAICohere and Hugging Face, there’s a crop of vendors building services on top of open source AI models. Sitting somewhere in the middle is AI21 Labs, an Israeli company that developed a model — Jurassic-1 Jumbo, which is roughly the size of GPT-3 — and slowly built products around it, including an “AI-as-a-service” platform called AI21 Studio that lets customers create virtual assistants, chatbots, content moderation tools and more.

Investors sense an opportunity, evidently. Today, AI21 Labs closed a $64 million Series B round that values the company at $664 million. Led by Ahren Innovation Capital Fund with participation from Mobileye CEO and co-founder Amnon Shashua, Walden Catalyst, Pitango, TPY Capital and Mark Leslie, the tranche brings A21Labs’ total capital raised to $118.5 million.

Co-founder and CEO Ori Goshen said that the new money will be put toward R&D, particularly developing larger and more sophisticated language models, and recruiting talent. AI21 Labs currently has 120 employees and plans to hire around 50 more by the end of the year, defying the macroeconomic trend.

“Fortunately, the pandemic has positively impacted business — as more companies migrated to remote work, individuals needed to convey in written text what they would normally share verbally,” Goshen told TechCrunch in an email interview. “[Our] proprietary large language models’ core capabilities allow for the ingestion of massive amounts of corporate data use to do … custom content creation, summarization, and classification.”

AI21 Labs was co-founded in 2017 by Goshen, Shashua, and Stanford University professor Yoav Shoham. The company’s first product was Wordtune, an AI-powered writing aid meant to compete with Grammarly, which suggests rephrasing text wherever users type. AI21 Studio was released last August, along with a “pay-as-you-go” service that allows developers to apply for access to custom models fine-tuned on datasets unique to their requirements.  

AI21 Labs
AI21 Labs offers a range of tuning parameters to customize the output of its models. Image Credits: AI21 Labs

Within AI21 Studio, AI21 Labs’ Jurassic-1 family of models can be used for paraphrasing (like generating short product names from product description), extracting figures from text and labeling emails and notes by topic or category. The models can also  summarize content through a feature in Wordtune dubbed Wordtune Read, including snippets from articles, reports and PDF files.

Because they’re trained on large amounts of data from the internet, including social media, language models are capable of generating toxic and biased text based on similar language that they encountered during training. AI21 Labs’ models are no different; in early testing, one researcher was able to prompt them to say “people who love Jews are closed-minded.” While AI21 Labs requires customers to agree to a terms of use policy and usage guidelines, it hasn’t implemented filters for potentially toxic content generated by its APIs.

AI21 Labs, which says it manually reviews requests for fine-tuned models to combat abuse, has claimed that its models are “marginally less biased” than GPT-3.

Regardless, according to Goshen, the models have an advantage in that they’re augmented with external knowledge sources like Wikipedia. The latest version of AI21 Labs’ Jurassic-1 model, Jurassic-X, uses what Goshen calls a “modular reasoning knowledge system” to enhance its answers with “discrete reasoning experts” such as online calculators and currency converters. Jurassic-X can answer “nontrivial” math operations phrased in natural language as a result, Goshen says, as well as simplify “complex” questions that might trip up other language models.

Of course, it’s worth noting that AI21 Labs hasn’t commissioned a comparison of its Jurassic-X models with other commercial language models, so claims are all we have to go on.

The company’s questionable recent marketing stunt doesn’t instill enormous confidence. In June, AI21 Labs launched a chatbot modeled on the legal opinions of the late Supreme Court justice Ruth Bader Ginsburg that several AI technology experts characterized as misleading. Responding to the criticism, AI21 Labs said that the chatbot was “just an experiment” and admitted it can give inaccurate responses that should be taken “with a grain of salt.”

When asked, Goshen declined to disclose firm revenue figures or even estimates of growth. But he said that Studio has “hundreds” of paying clients and design partners — none of which he was willing to identify by name — in addition to over 10,000 users of its free plan, while Wordtune has “millions” of users.

Given the cost of training sophisticated models, there’s likely significant investor pressure to expand. AI21 Labs’ own research pegs expenses for developing a text-generating model with 1.5 billion parameters (i.e., variables that the model uses to generate and analyze text) at as much as $1.6 million. Jurassic-1 Jumbo contains 178 billion parameters. That’s not accounting for hosting costs to serve the models; AI21 Labs says it retains the services of “several” third-party cloud providers both in the U.S. and abroad.

“[There’s a lack] of market knowledge because the language model technology is so nascent and just starting to gain adoption,” Goshen said. “With the new funding, AI21 Labs will continue in its mission of building AI systems with an unprecedented capacity to understand and generate natural language.”

More TechCrunch

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

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

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

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

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

Veo can generate few-seconds-long 1080p video clips given a text prompt.

Google’s image-generating AI gets an upgrade

At Google I/O, Google announced upgrades to Gemini 1.5 Pro, including a bigger context window. .

Google’s generative AI can now analyze hours of video

The AI upgrade will make finding the right content more intuitive and less of a manual search process.

Google Photos introduces an AI search feature, Ask Photos

Apple released new data about anti-fraud measures related to its operation of the iOS App Store on Tuesday morning, trumpeting a claim that it stopped over $7 billion in “potentially…

Apple touts stopping $1.8B in App Store fraud last year in latest pitch to developers

Online travel agency Expedia is testing an AI assistant that bolsters features like search, itinerary building, trip planning, and real-time travel updates.

Expedia starts testing AI-powered features for search and travel planning