Paris-based startup Nabla is changing the healthcare game with its innovative AI copilot for doctors, having recently secured a hefty $24 million in Series B funding. This round was led by Cathay Innovation and ZEBOX Ventures. Let’s dive into what Nabla offers and why it’s making waves in the medical field.
Transforming Medical Documentation
Nabla has developed an AI assistant that acts as a silent partner for medical professionals. It’s not about replacing doctors but enhancing their work.
Tech at Work: The AI assistant uses speech-to-text technology to transcribe doctor-patient conversations, highlight key data points, and generate detailed medical reports in minutes.
Customization and Storage: Reports are tailored to doctors’ needs and stored locally on the computer, making them easily accessible and exportable to electronic health record systems (EHRs).
Focus on Data Processing, Not Storing
Nabla’s approach to data is unique. They prioritize processing over storing. This means:
Privacy First: Audio and medical notes aren’t stored on servers without clear consent from both doctor and patient.
Correcting Errors: Doctors have the option to share medical notes with Nabla for transcription error correction, ensuring accuracy.
Impact on Healthcare
Nabla’s AI copilot is more than just a tool; it’s a time-saver for doctors. By handling administrative tasks, it lets medical professionals focus more on patient care.
Nabla’s Reach and Future Goals
Usage and Customers: The AI copilot is already in use by thousands of doctors, particularly in the U.S., following its rollout across Permanente Medical Group.
Long-Term Vision: While Nabla eyes FDA-approved clinical decision support, they remain committed to keeping physicians integral to healthcare.
The Bottom Line
Nabla’s AI assistant is a testament to how AI can work alongside professionals, not replace them. With the latest funding, Nabla is ready to change the way doctors use technology. They’re doing this while strictly following privacy and data rules. This is just the beginning of AI’s journey in enhancing healthcare efficiency and patient care. 🚀💡🏥
Let’s dive into some exciting news from the world of medicine and AI. A team of researchers just made a big leap in the fight against antibiotic resistance, and it’s all thanks to deep learning.
The Challenge: Antibiotic resistance is a massive problem. We’re running out of effective drugs to fight infections.
The Solution: Researchers used deep learning to find a whole new class of antibiotics. Think of it as AI digging through tons of data to find hidden treasure – but in this case, the treasure is life-saving drugs.
How Did They Do It?
Smart AI Tech: The team developed an AI method that’s all about understanding chemical structures. It’s like teaching AI to recognize patterns that make good antibiotics.
Testing and Results: They tested this AI on loads of different compounds. The goal? Find ones that kill bacteria but are safe for us.
Success!: And guess what? They nailed it. The AI found new antibiotics that are powerful against bacteria and gentle on human cells.
Why This Matters
A Step Forward: This isn’t just cool science – it’s a huge step in solving the antibiotic crisis.
AI’s Power: It shows how AI can do more than just play games or drive cars. It can actually help save lives.
Explainable AI: The best part? This AI isn’t a black box. Researchers can actually understand why and how it picks certain compounds. Transparency for the win!
The Big Picture
This research isn’t just a win for science; it’s a win for all of us. It’s about using technology to make the world a healthier place. And that’s something to be excited about!
Welcome to “Last Week in AI” your go-to source for the latest and most intriguing developments in technology and AI. This week, we’re diving into some breakthroughs and surprises that are reshaping the industry. From the innovative Fast Feedforward architecture in AI to the latest in video editing AI technology, we’ve got you covered with the stories that matter.
What is FFF?: Imagine a neural network, the brain of AI, but super-efficient. That’s FFF. It’s a fresh take on building AI that does its job using way less computing power. Think of it as a brainy eco-friendly car of the AI world.
Beating the Competition: There’s this other AI tech called mixture-of-experts networks. They’re cool, but FFF is like the new kid on the block showing them up. It’s quicker, more efficient, and gets to the right answers faster.
FFF’s Secret Sauce: Two big things here. First, it’s got this thing called noiseless conditional execution – it’s really good at ignoring useless data. Plus, it doesn’t need tons of neurons to make smart guesses. So, you don’t need a beast of a computer to run complex AI.
Why’s this a big deal? If you’re into AI, FFF could change the game. It means running smart AI models could get easier and cheaper. We’re talking better chatbots, sharper weather forecasts, and so much more. Bottom line: FFF is about doing a lot more with a lot less, and that’s huge for tech everywhere.
ChatGPT’s first year? Total game-changer in the tech world. Let’s dive into the highlights:
Massive App Success: The ChatGPT apps for phones are a hit. Launched in 2023, they’re a new way for folks to chat with AI. With over 110 million installs and nearly $30 million in the bank, it’s clear people are loving it.
Huge, But Not the Biggest: While it’s raking in cash, especially with its $19.99 monthly subscription, ChatGPT’s not the top dog in the chatbot revenue game. Still, it’s setting records with downloads and has a strong market position.
Who’s Using ChatGPT?: Turns out, it’s mostly younger guys, with a big chunk of users from the U.S. It’s got a whopping 180.5 million users globally, including big names in business.
So, what’s next? Expect even more growth. This first year’s just the beginning. ChatGPT is reshaping how we use AI in daily life and there’s a lot more to come!
GPT-4’s making big waves in healthcare, especially in radiology. It’s all about making things faster and more accurate. Let’s break it down:
Speed and Precision in Reporting: GPT-4’s churning out radiology reports way quicker than before, but still keeping them spot on. Studies show it’s making things more uniform and efficient.
Acing Comparisons and Exams: When pitted against traditional radiologist reports, GPT-4’s not only matching up but also being more concise. It’s even nailing medical exams, showing it’s not just fast, but also smart.
Project MAIRA and Beyond: This project’s all about pushing GPT-4 to its limits in radiology, aiming to make doctors’ lives easier and improve how patients get involved in their own care.
But hey, it’s not all smooth sailing. There’s a bunch of ethical stuff and clinical trials to figure out to make sure GPT-4 fits into healthcare the right way. Bottom line? GPT-4’s set to seriously shake up how we do medicine. Stay tuned for what’s next!
OpenAI’s GPT Store Launch Delayed to Next Year
OpenAI’s got some news: their much-anticipated GPT store’s launch is pushed to next year. Here’s what’s up:
Unexpected Hold-Up: Originally set for a December release, the GPT store’s debut is now delayed. OpenAI’s been dealing with some unexpected stuff, including a rollercoaster week with CEO Sam Altman’s ouster and return.
Store’s Purpose: The GPT store is meant to be a marketplace for users to sell and share GPTs they’ve created using OpenAI’s platform. It’s all about giving creators a chance to earn from their custom GPTs based on how much they’re used.
Still Making Progress: Despite the delay, the team’s not just sitting around. They’ve been tweaking and improving ChatGPT and their custom GPT platform, so there’s progress happening behind the scenes.
So, what does this mean? If you’re keen on AI and OpenAI’s work, this is a bit of a wait-and-see situation. The GPT store sounds like a big deal for AI enthusiasts and creators, but we’ll have to hang tight a bit longer. Stay tuned for more updates!
From Challenge to Innovation: The story starts with Demi Guo at an AI Film Festival. Despite having tools like Runway and Adobe, there were gaps. That’s when Guo, with Stanford’s Chenlin Meng, decided to create something better: an easy-to-use AI video generator. It clicked big time, attracting half a million users and major funding.
Pika 1.0: A Tech Breakthrough: Focused initially on anime, Pika’s AI model is a game-changer. It’s all about realistic videos and cool editing features, backed by some serious tech and influential investors.
Empowering Everyone to Create: Pika’s not just for pros. It’s for anyone who wants to tell a story through video. With its recent $55 million funding, the app’s making high-quality video creation simple for everyone.
Looking ahead, Pika plans to grow its team and maybe shift to a subscription model. The goal? To be the go-to tool for not just filmmakers, but anyone with a story to tell. In short, Pika’s reshaping the whole video editing landscape. Watch this space!
Adobe’s latest move? Snapping up Rephrase.ai, and it’s a big deal for video buffs and AI enthusiasts. Here’s why:
Rephrase.ai’s Cool Tech: Coming from IIT Bombay and IIT Roorkee brains, Rephrase.ai is no ordinary startup. Their AI turns text into slick videos. Just type, and poof, you’ve got a video. It’s like having a magic wand for video production.
Adobe’s AI Ambition: Adobe’s been diving deep into AI, remember Firefly and those nifty Photoshop tricks? Now, they’re adding Rephrase.ai to their arsenal, signaling a major push into generative AI.
A Game-Changer for Adobe Users: With Rephrase.ai in Adobe’s toolkit, think Premier Pro and After Effects turbocharged with AI. It’s set to revolutionize how we make videos, making it faster and more intuitive.
So, what does this mean for you? If you love making videos, marketing, or just dig tech, this is big news. Adobe’s gearing up to change the game in video production. Keep your eyes peeled – the future of creative video-making just got a lot more exciting!
Google
Google’s GNoMEAI Unlocks Millions of New Materials
Google DeepMind researchers say they’ve expanded the number of known stable materials by 380,000. Some could be useful for everything from batteries to superconductors—if they can make them.
GNoME’s Big Discovery: A deep learning tool named GNoME (Graph Networks for Materials Exploration) has hit a goldmine. It’s predicted the stability of 2.2 million new crystals, including a staggering 380,000 that are stable enough to maybe power future tech.
AI’s Role in Material Science: This isn’t just about numbers. It’s a game changer in how we find and make new materials. Using AI, scientists can now speed up the discovery and even predict which materials could actually work in the real world.
Sharing the Wealth: The cool part? The data on these 380,000 stable materials is now out there for the research community. This could lead to greener technologies and a whole new approach to material science.
Why’s this exciting? If you’re into tech, science, or the environment, this is big news. AI’s not just about robots and chatbots; it’s reshaping how we discover and develop materials that could change our world. Keep an eye on this – the future’s looking bright (and possibly greener) thanks to AI!
AGI
OpenAI’s Alleged Secret AI Project Q* Stirs Controversy
There’s some intriguing buzz around OpenAI and its CEO, Sam Altman. Here’s the story:
Secret AI Project ‘Q*’: Reports are swirling about a hush-hush AI system at OpenAI named ‘Q*’. Employees are saying it’s so advanced, it can ace math tests and think critically. But the kicker? They’re worried it’s getting too powerful, and their concerns might not be taken seriously.
Altman’s Reaction: When asked, Altman didn’t shoot down the existence of Q*. He referred to it as an “unfortunate leak.” This has only added more fuel to the fire about what Q* really is.
Safety Concerns in AI: The whole Q* saga taps into a bigger fear about Artificial General Intelligence (AGI) – the kind of AI that could outsmart humans. Experts have been flagging this for a while, worried about potential risks.
Now, it’s important to note that all this is based on claims and hasn’t been confirmed by OpenAI. But it sure has got people talking. Could this be why Altman briefly lost his CEO gig? For now, it’s a mix of speculation and concern in the AI community. Stay tuned as more unfolds.
Nvidia’s CEO Foresees AGI Within Five Years
Nvidia’s big boss, Jensen Huang, has made a bold prediction: Artificial General Intelligence (AGI) could be here in the next five years. Here’s the lowdown:
AGI: Smart as Humans?: Huang defines AGI as tech that’s pretty much as smart as a regular person. While AI’s been advancing like crazy, it’s not yet at the level of complex human smarts.
Concerns in the Tech World: It’s not all excitement, though. Big names in AI, like OpenAI’s Ilya Sutskever and investor Ian Hogarth, are worried about risks. Think fake news, cyberattacks, and scary AI weapons. There’s even talk about AGI possibly threatening humanity if it’s not kept in check.
Tech Leaders Agree: Huang’s not alone in his thinking. Other tech hotshots, like John Carmack of Meta and DeepMind’s Demis Hassabis, also believe powerful AI is just around the corner.
So, what’s the big picture? If you’re into tech and AI, this is huge. AGI could change everything, but it’s got some big names thinking hard about the risks. It’s a mix of excitement and caution as we head into a future where AI might be as smart as us. Stay tuned!
Elon Musk Forecasts AGI Breakthrough in Three Years
Elon Musk, always making headlines, has a fresh take on AI’s future. Here’s what you need to know:
BREAKING: Elon Musk has just stated at the DealBook summit that artificial general intelligence (AGI) is only about three years away.
He also stated that he thinks that Open AI discovered something possibly dangerous or alarming and that is why Sam Altman was fired.
Musk’s Bold AGI Prediction: In a chat with CNBC’s Andrew Sorkin, Musk dropped a bombshell. He thinks Artificial General Intelligence (AGI) is less than three years away. And not just any AGI, but one that could outdo the smartest humans in everything from writing novels to inventing new tech.
Mixed Reactions: As expected, Musk’s prediction has stirred up the pot. There’s a blend of excitement and skepticism in the air. Some folks are pumped about the possibilities, while others are raising their eyebrows.
Ripple Effect on Advertisers: Musk’s statement didn’t just stop at AI talk. It’s had a real-world impact, with many of his supporters urging boycotts against companies like Disney and Apple, who’ve pulled their ads from his ventures.
So, what’s the takeaway? Whether you’re a tech enthusiast, an AI skeptic, or just following the latest from Musk, this is a conversation starter. AGI in three years? It’s a bold claim and one that’s keeping the tech world buzzing. Let’s see how it unfolds!
Final Thoughts
That wraps up this week’s edition of “Last Week in AI.” These stories highlight the continuous and rapid evolution of AI and technology, impacting everything from healthcare to video editing. Stay tuned for more updates and insights next week as we continue to explore the cutting edge of tech and AI innovations.
If you missed last weeks update, you can check it out here.
The advent of GPT-4 marks a transformative era in radiology, showcasing how AI can reshape healthcare. We’re witnessing a significant leap from traditional practices to a future where AI aids in crucial medical domains.
Efficiency and Accuracy: GPT-4 has demonstrated its ability to generate radiology reports at a more efficient rate without compromising accuracy. Studies commissioned by the National Institutes of Health (NIH) reveal its potential in standardizing and optimizing radiology reporting.
Comparative Analysis: Researchers compared GPT-4 generated reports with those of radiologists, focusing on clarity, structure, and content. The AI-generated reports were not only similar in content but also more efficient, using fewer words and characters, thus enhancing communication in clinical practice.
Advancements in Medical Domains
Benchmark Performance: GPT-4’s impressive performance in medical competency exams and potential utility in consultations indicate a promising outlook for healthcare innovation.
Structured Reports: Structuring radiology reports is another area where GPT-4 excels. This structuring improves the standardization of disease descriptions, making them more interpretable and searchable for healthcare providers and research.
Project MAIRA: Multimodal AI in Radiology
Comprehensive Investigation: Project MAIRA focuses on a specialized large multimodal model for radiology report generation, exploring GPT-4’s boundaries in this field.
Impact on Radiologist Workflow: GPT-4’s introduction into radiology aims to enhance workflow efficiency and patient engagement, creating a future where patients and healthcare providers benefit from AI’s precision and clarity.
The Future Outlook
Patient Engagement and Education: Translating medical reports into more empathetic and understandable formats is a step forward in revolutionizing patient engagement and education.
Ethical Considerations and Clinical Trials: While GPT-4’s potential is vast, its implementation in clinical environments requires ethical considerations and extensive research through clinical trials to ensure responsible integration in healthcare.
All in All
GPT-4 is changing the game in healthcare. It’s bringing AI to radiology, blending it with what doctors know to make patient care safer and more on point. Everyone’s buzzing about how GPT-4 is going to shake things up in healthcare big time. We’re right at the edge of something super exciting in medicine!
Let’s dive into the latest in the world of AI: OpenAI’s leadership updates, xAI’s new chatbot, Google’s AI advancements, PANDA’s healthcare breakthrough, and the Genentech-NVIDIA partnership. Discover how these developments are transforming technology.
OpenAI
Sam Altman Reinstated as OpenAI CEO
Sam Altman is back as CEO of OpenAI after a dramatic boardroom drama. The conflict, which saw former president Greg Brockman resign and then return, ended with an agreement for Altman to lead again. The new board includes Bret Taylor, Larry Summers, and Adam D’Angelo, with D’Angelo representing the old board. They’re tasked with forming a larger, nine-person board to stabilize governance. Microsoft, a major investor, seeks a seat on this expanded board.
Leadership Reinstated: Altman’s return, alongside Brockman, signifies a resolution to the internal power struggle.
Board Restructuring: A new, smaller board will create a larger one for better governance, involving key stakeholders like Microsoft.
Future Stability: This change aims to ensure stability and focus on OpenAI’s mission, with investigations into the saga planned.
This shake-up highlights the challenges in managing fast-growing tech companies like OpenAI. It underscores the importance of stable leadership and governance in such influential organizations. For users and investors, this means a return to a focused approach towards advancing AI technology under familiar leadership.
OpenAI’s New AI Breakthrough Raises Safety Concerns
OpenAI, led by chief scientist Ilya Sutskever, achieved a major technical advance in AI model development. CEO Sam Altman hailed it as a significant push in AI discovery. Yet, there’s internal concern about safely commercializing these advanced models.
Technical Milestone: OpenAI’s new advancement marks a significant leap in AI capabilities.
Leadership’s Vision: Sam Altman sees this development as a major push towards greater discovery in AI.
Safety Concerns: Some staff members are worried about the risks and lack of sufficient safeguards for these more powerful AI models.
OpenAI’s advancement marks a leap in AI technology, raising questions about balancing innovation with safety and ethics in AI development. This underscores the need for careful management and ethical standards in powerful AI technologies.
OpenAI Researchers Warn of Potential Threats
OpenAI researchers have raised alarms to the board about a potentially dangerous new AI discovery. This concern was expressed before CEO Sam Altman was ousted. They warned against quickly selling this technology, especially the AI algorithm Q*, which might lead to AGI (artificial general intelligence). This algorithm can solve complex math problems. Their worries highlight the need for ethical and safe AI development.
AI Breakthrough: The AI algorithm Q* represents a significant advancement, potentially leading to AGI.
Ethical Concerns: Researchers are worried about the risks and ethical implications of commercializing such powerful AI too quickly.
Safety and Oversight: The letter stresses the need for careful, responsible development and use of advanced AI.
The situation at OpenAI shows the tricky task of mixing tech growth with ethics and safety. Researchers’ concerns point out the need for careful, controlled AI development, especially with game-changing technologies. This issue affects the whole tech world and society in responsibly using advanced AI.
ChatGPT Voice
ChatGPT Voice rolled out for all free users. Give it a try — totally changes the ChatGPT experience: https://t.co/DgzqLlDNYF
Inflection AI’s new ‘Inflection-2‘ model beats Google and Meta, rivaling GPT-4. CEO Mustafa Suleyman will upgrade their chatbot Pi with it. The model, promising major advancements, will be adapted to Pi’s style. The company prioritizes AI safety and avoids political topics, acknowledging the sector’s intense competition.
Innovative AI Model: Inflection-2 is poised to enhance Pi’s functionality, outshining models from tech giants like Google and Meta.
Integration and Scaling: Plans to integrate Inflection-2 into Pi promise significant improvements in chatbot interactions.
Commitment to Safety and Ethics: Inflection AI emphasizes responsible AI use, steering clear of controversial topics and political activities.
Inflection AI’s work marks a big leap in AI and chatbot tech, showing fast innovation. Adding Inflection-2 to Pi may create new benchmarks in conversational AI, proving small companies can excel in advanced tech. Their focus on AI safety and ethics reflects the industry’s shift towards responsible AI use.
Anthropic’s Claude 2.1
Claude 2.1 is a new AI model enhancing business capabilities with a large 200K token context, better accuracy, and a ‘tool use’ feature for integrating with business processes. It’s available via API on claude.ai, with special features for Pro users. This update aims to improve cost efficiency and precision in enterprise AI.
Extended Context Window: Allows handling of extensive content, enhancing Claude’s functionality in complex tasks.
Improved Accuracy: With reduced false statements, the model becomes more reliable for various AI applications.
Tool Use Feature: Enhances Claude’s integration with existing business systems, expanding its practical use.
Claude 2.1 is a major step in business AI, offering more powerful, accurate, and versatile tools. It tackles AI reliability and integration challenges, making it useful for diverse business operations. Its emphasis on cost efficiency and precision shows how AI solutions are evolving to meet modern business needs.
Claude 2.1 (200K Tokens) – Pressure Testing Long Context Recall
We all love increasing context lengths – but what's performance like?
Anthropic reached out with early access to Claude 2.1 so I repeated the “needle in a haystack” analysis I did on GPT-4
Elon Musk’s xAI is introducing Grok, a new chatbot, to its X Premium+ subscribers. Grok, distinct in personality and featuring real-time knowledge access via the X platform, is designed to enhance user experience. It’s trained on a database similar to ChatGPT and Meta’s Llama 2, and will perform real-time web searches for up-to-date information on various topics.
xclusive Chatbot Launch: Grok will be available to Premium+ subscribers, highlighting its unique features and personality.
Real-Time Knowledge Access: Grok’s integration with X platform offers up-to-date information, enhancing user interaction.
Amidst Industry Turbulence: The launch coincides with challenges at X and recent events at rival AI firm OpenAI.
xAI’s release of Grok is a key strategy in the AI chatbot market. Grok’s unique personality and real-time knowledge features aim to raise chatbot standards, providing users with dynamic, informed interactions. This launch shows the AI industry’s continuous innovation and competition to attract and retain users.
Google’s Bard AI Gains Video Summarization Skill, Sparks Creator Concerns
Google’s Bard AI chatbot now can analyze YouTube videos, extracting key details like recipe ingredients without playing the video. This skill was demonstrated with a recipe for an Espresso Martini. However, this feature, which is part of an opt-in Labs experience, could impact content creators by allowing users to skip watching videos, potentially affecting creators’ earnings.
Advanced Video Analysis: Bard’s new capability to summarize video content enhances user convenience.
Impact on YouTube Creators: This feature might reduce views and engagement, affecting creators’ revenue.
Balancing Technology and Creator Rights: The integration of this tool into YouTube raises questions about ensuring fair value for creators.
Bard’s latest update illustrates the evolving capabilities of AI in media consumption, making content more accessible. However, it also highlights the need for a balance between technological advancements and the rights and earnings of content creators. Google’s response to these concerns will be crucial in shaping the future relationship between AI tools and digital content creators.
Google Bard now lets you chat with YouTube videos using AI.
Before it could find relevant videos but now it understands them.
PANDA: AI for Accurate Pancreatic Cancer Detection
A study in Nature Medicine presents PANDA, a deep learning tool for detecting pancreatic lesions using non-contrast CT scans. In tests with over 6,000 patients from 10 centers, PANDA exceeded average radiologist performance, showing high accuracy (AUC of 0.986–0.996) in identifying pancreatic ductal adenocarcinoma (PDAC). Further validation with over 20,000 patients revealed 92.9% sensitivity and 99.9% specificity. PANDA also equaled contrast-enhanced CT scans in distinguishing pancreatic lesion types. This tool could significantly aid in early pancreatic cancer detection, potentially improving patient survival.
Exceptional Accuracy: PANDA shows high accuracy in detecting pancreatic lesions, outperforming radiologists.
Large-Scale Screening Potential: Its efficiency in a multi-center study indicates its suitability for widespread screening.
Early Detection Benefits: Early detection of PDAC using PANDA could greatly improve patient outcomes.
PANDA represents a major advancement in medical AI, offering a more effective way to screen for pancreatic cancer. Its high accuracy and potential for large-scale implementation could lead to earlier diagnosis and better survival rates for patients, showcasing the impactful role of AI in healthcare diagnostics.
Genentech and NVIDIA Partner to Accelerate Drug Discovery with AI
Genentech and NVIDIA are collaborating to advance medicine development with AI. They’re enhancing Genentech’s algorithms using NVIDIA’s supercomputing and BioNeMo platform, aiming to speed up and improve drug discovery. This partnership is set to boost efficiency in scientific innovation and drug development.
Optimized Drug Discovery: Genentech’s AI models will be enhanced for faster, more successful drug development.
AI and Cloud Integration: Leveraging NVIDIA’s AI supercomputing and BioNeMo for scalable model customization.
Mutual Expertise Benefit: Collaboration provides NVIDIA with insights to improve AI tools for the biotech industry.
This collaboration marks a significant advance in integrating AI with biotech, potentially transforming how new medicines are discovered and developed. By combining Genentech’s drug discovery expertise with NVIDIA’s AI and computational prowess, the partnership aims to make the drug development process more efficient and effective, promising faster progress in medical innovation.
The AI world is rapidly evolving, from OpenAI’s changes to innovative healthcare tools. These developments demonstrate AI’s growing impact on technology and industries, underscoring its exciting future.
