( September 17, 2025, 08:42 GMT | Official Statement) — MLex Summary: South Korea’s Ministry of SMEs and Startups said Wednesday it will fully support entrepreneurs in artificial intelligence and other deep-tech fields, while announcing a 13.5 trillion won ($9.8 billion) program to help startups grow into unicorns. The program, to be run alongside the government’s 150 trillion won National Growth Fund, will give “promising companies” investment tailored to their growth stages. The ministry also said the government will build a cross-ministerial support system for startups in key technology sectors including AI, defense and climate tech. To back their overseas expansion, a “startup and venture campus” will also be set up in Silicon Valley to provide integrated services that help startups settle and grow abroad, the ministry added.
The statement, in Korean, is attached….
Tools & Platforms
Flatiron Health study shows AI can match human experts in tracking cancer progression
AI can track cancer progression via unstructured notes in the EHR: ©Infinite Flow – stock.adobe.com
Flatiron Health announced research findings showing that large language models (LLMs) can accurately and efficiently extract cancer progression data from unstructured electronic health records, potentially boosting oncology research and care.
The study, presented at a recent conference on artificial intelligence in cancer research, found that AI tools achieved F1 scores comparable to those of expert human abstractors across 14 cancer types. The LLM used in the study, provided by Anthropic, also generated nearly identical estimates of real-world progression-free survival, according to the researchers.
“AI and machine learning are fundamentally transforming how we generate and use real-world evidence in oncology,” said Stephanie Reisinger, senior vice president and general manager of real-world evidence at Flatiron Health. “This research exemplifies how Flatiron is harnessing AI and multimodal data to unlock new insights from oncology real-world data—accelerating clinical research, improving patient outcomes, and setting a new standard for evidence generation in cancer care.”
To validate the quality of the AI-extracted data, Flatiron used its VALID (Validation of Accuracy for LLM/ML-Extracted Information and Data) framework, comparing the AI’s performance to both a primary and duplicate expert human abstractor.
“Scalable, high-quality extraction of such an important and complex endpoint like progression will open new doors for novel research, predictive modeling, and more personalized patient care,” said Aaron B. Cohen, lead author and practicing oncologist at Bellevue Hospital in New York City.
Flatiron also presented work on AI fairness in data extraction, emphasizing the need for bias evaluation as AI continues to play a larger role in health care.
AI’s expanding role in oncology data analysis
As artificial intelligence continues to evolve, its application in oncology has moved beyond imaging and diagnostics into the complex world of electronic health records. Traditionally, extracting useful clinical insights from unstructured EHR data—such as physician notes, pathology reports, and treatment narratives—has required manual review by trained professionals. This process is not only time-consuming and expensive but also difficult to scale.
Recent advances in natural language processing and machine learning have made it possible to automate the extraction of clinically meaningful data with a high degree of accuracy. These tools can now identify cancer progression events, treatment responses, comorbidities, and adverse events buried within unstructured text. Importantly, they do so at a speed and scale that far exceeds human capacity.
Key to these advances is the development of validation frameworks and performance metrics that allow AI-generated data to be compared directly with human-extracted data. The focus is shifting toward not only accuracy but also fairness, transparency, and reproducibility. Researchers are increasingly incorporating bias detection protocols to ensure these systems work equally well across diverse populations.
Beyond data extraction, AI is also being used to synthesize multimodal datasets—including genomic, clinical, and imaging data—to drive predictive modeling, identify candidates for clinical trials, and inform personalized treatment pathways.
As these systems mature, experts believe they will become integral to real-world evidence generation, allowing researchers and clinicians to answer complex clinical questions more quickly, identify emerging treatment patterns, and ultimately improve outcomes for cancer patients worldwide.
“Employees who don’t really understand the goal they’re after will spin in circles not knowing when they should just call it done or step away,” Farmer says. “The enemy of good is perfect, and LLMs make us feel like if we just tweak that last prompt a little bit, we’ll get there.”
Agents of doom
Observers see two versions of doomprompting, with one example being an individual’s interactions with an LLM or another AI tool. This scenario can play out in a nonwork situation, but it can also happen during office hours, with an employee repeatedly tweaking the outputs on, for example, an AI-generated email, line of code, or research query.
The second type of doom prompting is emerging as organizations adopt AI agents, says Jayesh Govindarajan, executive vice president of AI at Salesforce. In this scenario, an IT team continuously tweaks an agent to find minor improvements in its output.
Since the July rollout of Ally’s proprietary artificial intelligence platform, the breadth of use is what’s surprised Sathish Muthukrishnan, the bank’s chief information, data and digital officer.
“We have people in the sales force that are using it, people in the operations side, customer care associates using it; obviously, folks in the technology side; marketing; our risk control partners, risk compliance; audit, privacy – they’re all big users of it,” said Muthukrishnan, who’s been in his role at the digital bank since 2019.
The Detroit-based lender gave its 10,000 employees access to Ally.ai two months ago, after testing it with a smaller group for more than a year. About 400,000 prompts have been submitted to the platform, and adoption is at about 50%.
The bank wants employees to use the platform, which was built in-house, to handle tasks such as drafting emails and proofreading copy, to free up their time for other projects.
When asked how AI might affect the company’s headcount, Muthukrishnan said it’s set to “have a meaningful impact on the business outcomes.”
Ally has “ambitious” growth plans, so for the company to generate more revenue while maintaining current spending levels, “technology and AI become critical,” Muthukrishnan said in a recent interview with Banking Dive. “That’s both driving efficiency and effectiveness. It’s not just efficiency of cost; it’s efficiency of speed.”
Editor’s note: This interview has been edited for clarity and brevity.
BANKING DIVE: Where does Ally go from here with AI?
SATHISH MUTHUKRISHNAN: Since the launch, there is tremendous demand and a lot of use cases coming our way. Now, let’s turn the tables and see how we can identify use cases that are harder to solve on the business side, and how do we bring that to the forefront?
With the pace at which technology is evolving, something that seems impossible, something that seems super hard to solve right now, we will be able to solve in a few months. So we want to tackle those hard problems now, and we want to do it collectively across the organization.
Our CEO has asked me to come and educate the entire executive committee on how we are advancing in AI, and we’re going to call it an executive committee AI day, and it’s just purely to set aside dedicated time, bring us all together, fully focused on AI. These are all busy people running big organizations, so there’s a little bit of pressure on making sure that I use their time efficiently. But we’re going to talk about what are the things that we can collectively solve for the company. We have thoughtfully rolled out AI, and there is interest across the company, but we need to bring the company along.
How has Ally’s AI governance approach evolved since implementation?
It might sound like a cliche, but we focus on doing simple things savagely well. Things that are simple – having risk controls, having data protection, having access controls – can be cast aside because you see the shinier object.
For us, to have an AI working group, then having an AI governance steering council, then having an enterprise-level committee, then the board – having this many levels of governance to ensure that AI is scaled safely and responsibly is super critical. We did the hard work ahead of time, we have exercised this governance muscle extremely well, and people have gotten used to it.
How do you see the role of AI agents evolving at Ally in the coming years?
Agentic AI allows you to look at the complicated paths, complicated processes, and allows you to digitize that. It’s still in an experimental stage for us.
For example, all applications in our tech ecosystem have observability. If there is an issue, we want to be the first to find out, before the customer finds out, or our business partner finds out. So a ton of alerts come our way. If I have to process those alerts, but not increase my headcount as I’m increasing the number of customers, I’m looking at agentic AI to do that. The usage of digital has doubled in the last four years by our customers, but the cost of serving them has gone down. That’s because of the introduction of new technology.
If you want somebody to reset your password, that could be agentic AI that does that internally. Those are some of the experiments that we are doing; nothing that is in production or at scale yet.
What are your biggest concerns, as the bank’s tech chief?
No. 1 is cyber. It’s top of mind. Not just because of what is happening today, but how easy it is for cyber criminals to get access to technology and create more sophisticated cyber attacks. But as a regulated institution, you have to watch how technology is being adopted.
No. 2 is the pace at which the technology is changing. It weighs on me, because it’s a constant learning process for the organization, on top of delivering at pace. So your learning agility and your execution agility always has to be, like, in the fifth gear, and that’s hard. And these cyber events are like sharp turns or speed bumps on the road while you’re running a marathon at sprint speed.
And No. 3 is, how do you educate and bring the organization along? Technology or AI is not here to replace anybody; it is to make sure that you take away the mundane part of the job and make them even more effective, so they can serve our customers better. So how do you do that, how do you bring everybody along? And that is a patience game. But at the same time, the speed at which the technology is changing, you can’t be patient. So it’s a partnership with them, to bring them along.
How do you think about the future of your role?
Today, everybody can be a CIO because of ChatGPT. You would walk into a room and you felt like you had knowledge that nobody else did. But as you’re talking, people can be on their phone verifying your answer or getting a better answer, because ChatGPT is the better CIO, whether we like it or not.
On the other hand, thinking about the art of the possible, using technology as a tool to re-underwrite, re-imagine your business and asking the difficult questions of, are you serving your purpose? Are you serving your customers right? Do you have the right talent and capabilities to serve the customers? Can you do it differently? Can you do it faster? Can you do it with better technology? Can you enhance the user experience?
It’s up to the technology leader to figure out, where do you want to be? Do you want to be in a defensive mode, or do you want to be in an offensive mode, partnering with business leaders and figuring out, what is the next version of the product going to be?
Prepare for tomorrow’s regulatory change, today
MLex identifies risk to business wherever it emerges, with specialist reporters across the globe providing exclusive news and deep-dive analysis on the proposals, probes, enforcement actions and rulings that matter to your organization and clients, now and in the longer term.
Know what others in the room don’t, with features including:
- Daily newsletters for Antitrust, M&A, Trade, Data Privacy & Security, Technology, AI and more
- Custom alerts on specific filters including geographies, industries, topics and companies to suit your practice needs
- Predictive analysis from expert journalists across North America, the UK and Europe, Latin America and Asia-Pacific
- Curated case files bringing together news, analysis and source documents in a single timeline
Experience MLex today with a 14-day free trial.
-
Business3 weeks agoThe Guardian view on Trump and the Fed: independence is no substitute for accountability | Editorial
-
Tools & Platforms1 month ago
Building Trust in Military AI Starts with Opening the Black Box – War on the Rocks
-
Ethics & Policy2 months agoSDAIA Supports Saudi Arabia’s Leadership in Shaping Global AI Ethics, Policy, and Research – وكالة الأنباء السعودية
-
Events & Conferences4 months agoJourney to 1000 models: Scaling Instagram’s recommendation system
-
Jobs & Careers3 months agoMumbai-based Perplexity Alternative Has 60k+ Users Without Funding
-
Podcasts & Talks2 months agoHappy 4th of July! 🎆 Made with Veo 3 in Gemini
-
Education2 months agoMacron says UK and France have duty to tackle illegal migration ‘with humanity, solidarity and firmness’ – UK politics live | Politics
-
Education3 months agoVEX Robotics launches AI-powered classroom robotics system
-
Podcasts & Talks2 months agoOpenAI 🤝 @teamganassi
-
Funding & Business3 months agoKayak and Expedia race to build AI travel agents that turn social posts into itineraries