Connect with us

Business

There’s a Looming AI Data Shortage. Google Researchers Have a New Fix.

Published

on


Google DeepMind researchers have an idea for how to solve the AI data drought, and it might involve your Social Security number.

The large language models powering AI require vast amounts of training data pulled from webpages, books, and other sources. When it comes to text specifically, the amount of data on the web considered fair game for training AI models is being scraped faster than new data is being created.

However, a large portion of the data isn’t used because it’s deemed toxic, inaccurate, or it contains personally identifiable information.

In a newly published paper, a group of Google DeepMind researchers claim to have found a way to clean up this data and make it usable for training, which they claim could be a “powerful tool” for scaling up frontier models.

They refer to the idea as Generative Data Refinement, or GDR. The method uses pretrained generative models to rewrite the unusable data, effectively purifying it so it can be safely trained on. It’s not clear if this is a technique Google is using for its Gemini models.

Minqi Jiang, one of the paper’s researchers who has since left the company to Meta, told Business Insider that a lot of AI labs are leaving usable training data on the table because it’s intermingled with bad data. For example, if there’s a document on the web that contains something considered unusable, such as someone’s phone number or an incorrect fact, labs will often discard the entire thing.

“So you essentially lose all those tokens inside of that document, even if it was a small single line that contained some personally identifying information,” said Jiang. Tokens are the units of data, processed by AI, which make up words within text.

The authors give an example of raw data that included someone’s Social Security number or information that may soon be out of date (“the incoming CEO is…”). In these instances, the GDR would swap or remove the numbers, ignore the information that risks becoming obsolete, and retain the remainder of usable data.

The paper was written more than a year ago and was only published this month. A Google DeepMind spokesperson did not respond to a request for comment about whether the researcher’s work was being applied to the company’s AI models.

The authors’ findings could prove helpful for labs as the usable well of data runs dry. They cite a research paper from 2022 that predicted AI models could soak up all the human-generated text between 2026 and 2032. This prediction was based upon the amount of indexed web data, using statistics from Common Crawl, a project that continuously scrapes web pages and makes them openly available for AI labs to use.

For the GDR paper, the researchers performed a proof of concept by taking over one million lines of code and having human expert labelers annotate the data line by line. They then compared the results with the GDR method.

“It completely crushes the existing industry solutions being used for this kind of stuff,” said Jiang.

The authors also said their method is better than the use of synthetic data (data generated by AI models for the purpose of training themselves or other models), which has been a topic of exploration among AI labs. However, using synthetic data can degrade the quality of model output and, in some cases, lead to “model collapse.”

The authors compared the GDR data against synthetic data created by an LLM and discovered that their approach created a better dataset for training AI models.

They also said further testing could be conducted on other complicated types of data considered a no-go, such as copyrighted materials and personal data that is inferred across multiple documents rather than explicitly spelled out.

The paper has not been peer reviewed, said Jiang, adding that this is common in the tech industry and that all papers are reviewed internally.

The researchers only tested GDR on text and coding. Jiang said that it could also be tested on other modalities, such as video and audio. However, given the rate at which new videos are generated each day, they’re still providing a firehose of data for AI to train on.

“With video, you’re just going to have a lot more of it, just because there’s a constant stream of millions of hours of video generated each day,” said Jiang. “So I do think, going across new modalities beyond text, video, and images, we’re going to unlock a lot more data.”

Have something to share? Contact this reporter via email at hlangley@businessinsider.com or Signal at 628-228-1836. Use a personal email address and a non-work device; here’s our guide to sharing information securely.





Source link

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Business

CrowdStrike and Salesforce Partner to Secure the Future of AI-Powered Business

Published

on


CrowdStrike and Salesforce announced a new strategic partnership to enhance the security of AI agents and applications built on Agentforce and the Salesforce Platform. Through integrations between CrowdStrike Falcon®? Shield and Salesforce Security Center, Salesforce admins and security professionals will gain enhanced visibility, compliance support, and protection for mission-critical workflows – simplifying operations and uniting business and security teams on a shared foundation of trust in the agentic era.

The partnership also enables customers to access CrowdStrike’s agentic security analyst, Charlotte AI, through Agentforce for Security and use it to work directly alongside teammates in Slack, flagging potential threats and recommending actions in a conversational manner as any other employee would. As agents join the workforce, security teams must understand what they are doing, trace them back to their human creators, and prevent them from becoming over privileged or compromised. CrowdStrike and Salesforce are meeting this challenge by delivering the visibility and control needed to secure the future of AI-powered business.

Automatic Threat Containment: Automate response actions with Falcon®? Fusion – such as blocking risky access or disabling compromised agents – directly from Salesforce Security Center. Unified AI Agent Protection: Combine Falcon Shield, Falcon®?

Next-Gen Identity Security, and Falcon®? Cloud Security to deliver end-to-end control over Agentforce agents and applications. By bringing Charlotte AI into Slack through Agentforce for Security, CrowdStrike and Salesforce empower teams to quickly and efficiently handle security incidents without having to switch applications: Accelerated Incident Response: Instantly create dedicated incident channels in Slack to coordinate response; Conversational Threat Investigation: Use natural language to query Charlotte AI for immediate answers on threats, hosts, and data; Real-Time Remediation: Isolate compromised devices or take other response actions directly from Slack, ensuring swift containment.

Together, CrowdStrike and Salesforce deliver stronger protection and visibility for mission-critical workflows- enabling enterprises to embrace AI securely while building the foundation for future innovation. Availability: The Falcon Shield integration will be available from within the Salesforce Security Center and on the Salesforce AppExchange this year; Charlotte AI will be integrating into Slack via Agentforce for Security and available via the AgentExchange and Slack Marketplace this year.



Source link

Continue Reading

Business

Workday Signs Definitive Agreement to Acquire Sana

Published

on


Acquisition to Turn Workday into the New Front Door for Work

Acquisition Will Combine Sana’s AI-Powered Search, Agents, and Learning with Workday Context and Data to Power Proactive, Personalized, and Intelligent Employee Experiences

SAN FRANCISCO, Sept. 16, 2025 /PRNewswire/ — Workday Rising 2025 — Workday, Inc. (NASDAQ: WDAY), the enterprise AI platform for managing peoplemoney, and agents, has entered into a definitive agreement to acquire Sana, a leading AI company building the next generation of enterprise knowledge tools. Sana will power a new Workday experience—where knowledge, data, action, and learning come together as one and create the new front door for work.

Since its founding in 2016, Sana has been at the forefront of AI for work, developing intuitive tools that elevate humans with AI. Sana’s core products, Sana Learn and Sana Agents, have already served over one million users across hundreds of enterprises.

In addition to powering a new Workday experience, Sana will continue to develop Sana Learn and Sana Agents. As part of Workday, Sana will be able to accelerate its growth and deliver even more innovation to its customers at scale.

“Sana’s team, AI-native approach, and beautiful design perfectly align with our vision to reimagine the future of work,” said Gerrit Kazmaier, president, product and technology, Workday. “This will make Workday the new front door for work, delivering a proactive, personalized, and intelligent experience that unlocks unmatched AI capabilities for the workplace.”

“Our focus has always been on creating intuitive AI tools that improve how people learn and work,” said Joel Hellermark, founder and CEO of Sana. “I’m excited to bring these tools to 75 million Workday users and partner with Workday’s iconic team to launch a new era of superintelligence for work.”

The New Front Door for Work: A Reimagined Workday Experience 

With Sana, Workday will create the work experience of the future, where enterprise knowledge, data and actions converge into one. This will help people get their work done and empower employees with AI agents that can:

  • Find answers, information and files by instantly searching across a company’s most critical data sources, including Workday, Google Drive, SharePoint, and Office365.
  • Act proactively by anticipating needs, summarizing insights, and assisting with projects.
  • Create presentations, documents, and dashboards, even full learning courses, based on company knowledge.
  • Automate repetitive tasks and routine work by executing workflows end-to-end.

Leveraging Workday’s unique data and context around people and money—as well as a rich ecosystem of builders and partners—the employee experience will become personalized and proactive, better anticipating employee needs based on their role, team, and projects. For example, hiring managers will be able to generate tailored dashboards to monitor their live recruitment pipeline, automate the end-to-end performance review process, and receive proactive suggestions on onboarding new hires based on real-time performance data.

Unlocking a New Era of Enterprise AI 

Sana Agents extends enterprise AI beyond basic search and chat. With the platform’s no-code agent builder, users can create AI agents to automate repetitive tasks and act proactively on their behalf. These agents streamline workflows while helping ensure that every action remains secure and compliant with company policies through the Workday Agent System of Record. 

Existing customers are realizing significant tangible value from Sana Agents across various use cases. For instance, a leading American manufacturer achieved up to 95% time savings; a multinational industrial tech company achieved 90% productivity gains; and a global law firm saw over 60% time savings and 200% increased efficiency.

Elevating Talent Development with AI-Powered Learning

Sana is also a pioneer in applying AI to learning. Its AI-native learning platform, Sana Learn, combines learning management, content creation, course generation, and personalized tutoring through specialized learning agents. Sana Learn has already enabled hundreds of customers across industries to accelerate learning. For example, a global electric vehicle manufacturer boosted learning engagement by 275%; a leading European installation distributor with 7,500 employees cut course creation time from four months to four days; and a global fintech company went from three weeks to three hours for content creation.

Sana Learn will complement Workday Learning with hyper-personalized skill building capabilities and AI-native content creation at scale. Enhanced by AI-driven internal mobility with Workday Talent Optimization and HiredScore, this comprehensive learning suite will help employees build skills faster and help enable organizations to scale personalized learning experiences, supporting employee reskilling and upskilling initiatives. 

“Sana pioneered the world of intelligent agents and AI-native learning at scale,” said Josh Bersin, global industry analyst and CEO of The Josh Bersin Company and a Sana customer. “I think Sana’s AI agent and learning system gives Workday customers the opportunity to completely transform the way their employees learn, grow, and operate as super workers in this new age of AI.”

Details Regarding Proposed Acquisition of Sana

Under the terms of the definitive agreement, Workday will acquire all of the outstanding shares of Sana for approximately $1.1 billion

The transaction is expected to close in the fourth quarter of Workday’s fiscal year 2026, ending January 31, 2026, subject to the satisfaction of customary closing conditions. Allen & Company LLC is serving as financial advisor to Workday and Orrick is serving as its legal advisor. DLA Piper is serving as Sana’s legal advisor.

About Workday

Workday is the enterprise AI platform for managing people, money, and agents. Workday unifies HR and Finance on one intelligent platform with AI at the core to empower people at every level with the clarity, confidence, and insights they need to adapt quickly, make better decisions, and deliver outcomes that matter. Workday is used by more than 11,000 organizations around the world and across industries – from medium-sized businesses to more than 65% of the Fortune 500. For more information about Workday, visit workday.com.

About Sana

Sana is an AI company building the next generation of knowledge tools. Its products have served over a million users globally and are trusted by the likes of Merck and Polestar. To learn more about Sana, visit sanalabs.com.

Forward-Looking Statements 

This press release contains forward-looking statements related to Workday, Sana, and the acquisition of Sana by Workday. These forward-looking statements are based only on currently available information and Workday’s current beliefs, expectations, and assumptions. Because forward-looking statements relate to the future, they are subject to risks, uncertainties, assumptions, and changes in circumstances that are difficult to predict and many of which are outside of our control. If the risks materialize, assumptions prove incorrect, or we experience unexpected changes in circumstances, actual results could differ materially from the results implied by these forward-looking statements, and therefore you should not rely on any forward-looking statements. Forward looking statements in this communication include, among other things, statements about the potential benefits and effects of the proposed transaction; Workday’s plans, objectives, expectations, and intentions with respect to Sana’s business; and the anticipated timing of closing of the proposed transaction. Risks include, but are not limited to: (i) the risk that the transaction may not be completed in a timely manner or at all; (ii) failure to achieve the expected benefits of the transaction; (iii) Workday’s ability to deliver a new Workday experience, accelerate Sana’s growth, and implement its other plans, objectives, and expectations with respect to Sana’s business and technology; (iv) negative effects of the announcement or the consummation of the transaction on Workday’s business operations, operating results, or share price; (v) unanticipated expenses related to the acquisition; and (vi) other risks and factors described in our filings with the Securities and Exchange Commission (“SEC”), including our most recent report on Form 10-Q or Form 10-K and other reports that we have filed and will file with the SEC from time to time, which could cause actual results to vary from expectations. Workday assumes no obligation to, and does not currently intend to, update any such forward-looking statements after the date of this release.

© 2025 Workday, Inc. All rights reserved. Workday and the Workday logo are registered trademarks of Workday, Inc. All other brand and product names are trademarks or registered trademarks of their respective holders.

SOURCE Workday Inc.

For further information: Investor Relations, ir@workday.com; Media, media@workday.com



Source link

Continue Reading

Business

AI Company ServiceNow Takes Up to 200K SF With Stephen Ross in West Palm Beach – Commercial Observer

Published

on


AI company ServiceNow is taking up to 200,000 square feet at Stephen Ross’s 10 CityPlace development in Downtown West Palm Beach, Fla. 

The Santa Clara, Calif.-based company will become the anchor tenant of the 480,000-square-foot development, which remains under construction. ServiceNow, which recorded nearly $11 billion in revenue last year, runs a cloud-based platform that helps firms automate and manage digital workflows using artificial intelligence.

SEE ALSO: L.A. Office Market Shifts Create More Opportunities for a Certain Investor

The City of West Palm Beach and the State of Florida have approved $17 million in incentives for ServiceNow if it creates 856 jobs, WPTV reported

ServiceNow plans to open an innovation hub within the office, which is expected to open in 2028. Ross’s firm, Related Ross, is negotiating to receive about $700 million in construction financing to build the tower as well as another office building next door, 15 CityPlace

“West Palm Beach is the latest move in ServiceNow’s tradition of embracing bold economic developments across the country,” Bill McDermott, chairman and CEO of ServiceNow, said in a statement. “This will be a compelling magnet for talent, a strong engine for growth, and a dynamic hub for America’s AI leadership.” 

ServiceNow’s lease marks a win for Ross’s broader quest to turn West Palm Beach into a leading business hub. 360 Rosemary, the office building that Ross completed in 2021, has landed high-profile finance tenants such as Goldman Sachs, J.P. Morgan and Elliott Investment Management

Ross has also successfully lobbied Vanderbilt University to open a $520 million graduate campus, though construction has yet to commence.

Julia Echikson can be reached at jechikson@commercialobserver.com



Source link

Continue Reading

Trending