Good morning, itās Wednesday. Chinaās DeepSeek is slashing prices to challenge OpenAI, forcing Big Tech to rethink its strategy. An old economic paradox is making a comeback, and OpenAIās high-priced āOperatorā assistant shows promiseābut doesnāt quite live up to the hype just yet.
Plus, in our latest Forward Future Original, we explore how OpenAIās Deep Research is advancing AI-driven analysis and inching us closer to AGI.
šļø YOUR DAILY ROLLUP
Top Stories of the Day
š EU Sets AI Rules as Compliance Deadline Arrives
The European Commission has released guidance to help developers comply with the EU AI Act, which recently began enforcing bans on āunacceptable riskā AI applications. While not legally binding, the guidelines clarify prohibited uses and potential penalties, which can reach 7% of global turnover or ā¬35 million. The AI Actās rollout will continue in phases, with full enforcement still dependent on EU member states designating oversight bodies by August 2.
š AI Gets a Speed Boost with MITās SySTeC
MIT researchers have unveiled SySTeC, a compiler that optimizes AI computations by reducing redundancy in data processing. Unlike traditional methods that force developers to choose between sparsity and symmetry optimizations, SySTeC combines both, accelerating AI models by up to 30 times. This innovation simplifies coding, cuts computational costs, and lowers energy demands, making deep-learning models more efficient and accessible to researchers beyond the AI field.
š ByteDanceās Deepfake AI Makes Reality Hard to Trust
ByteDance has introduced OmniHuman-1, an advanced deepfake AI that creates hyper-realistic videos from a single image and audio clip. Unlike typical deepfakes, it seamlessly mimics facial and body movements, making detection even harder. While not publicly available yet, similar technologies spread fast, fueling misinformation and fraud. With AI-driven scams projected to cost the U.S. $40 billion by 2027, calls for regulation are growing.
š± Why AI Leaders Fear an Old Economic Paradox
Microsoftās Satya Nadella recently cited Jevons paradox to explain why cheaper AIālike DeepSeekās chatbotāmight boost demand instead of slashing profits. Originally about coal efficiency, the 160-year-old theory suggests that AI-driven productivity gains could create more jobs in fields like coding. However, history warns that automation doesnāt benefit everyoneājust like farmers displaced by agricultural tech, some industries may face major disruptions.
š° Muskās Allies Bet on AI to Cut Government Waste
Elon Muskās allies in the Trump administration are exploring AI-driven audits to reduce federal spending, with some agencies facing potential 50% budget cuts. A top official at the General Services Administration proposes using AI to centralize contract data and identify inefficiencies, with Musk claiming it could uncover $1 trillion in waste. While his influence in Washington expands, experts question whether AI can truly deliver such sweeping savings.
š¾ FORWARD FUTURE ORIGINAL
Deep Research & AGI
On February 2, 2025, OpenAI presented āDeep Research,ā a groundbreaking AI feature that has the potential to fundamentally change the way we collect and analyze complex information. This new capability enables ChatGPT to independently conduct multi-step research on the internet, accomplishing in minutes what would take a human many hours to do.
Deep Research is designed to create extensive and detailed reports by searching, analyzing and synthesizing hundreds of online sources. Powered by a version of the upcoming OpenAI o3 model (!), optimized for web browsing and data analysis, it can search and interpret massive amounts of text, images and PDFs on the internet. This ability makes it possible to collect, evaluate and consolidate information to create comprehensive reports at the level of a research analyst.
The introduction of Deep Research marks a significant advance in the development of AI agents capable of completing tasks independently. While earlier models such as GPT-4o were suitable for real-time conversations, Deep Research goes a step further by processing in-depth, domain-specific queries where depth and detail are crucial. This ability to synthesize knowledge is a prerequisite for creating new knowledge and thus represents an important milestone on the road to developing Artificial General Intelligence (AGI).
Masayoshi Son, probably the most important investor in OpenAI's Stargate project, has announced that AGI, which he believes will come very soon, will initially be used in Japanese companies. ā Continue reading here.
š„ COMPETITION
AI Underdog Finds a Competitive Edge
The Recap: Chinese AI model DeepSeek is shaking up the industry by offering companies a significantly cheaper alternative to OpenAIās models, leading European startups to adopt it in droves. While its long-term viability remains uncertain, its disruptive pricing has sparked competition, raising questions about cost, security, and geopolitical risks.
European startups like Novo AI and NetMind.AI have quickly switched to DeepSeek, with some claiming savings of up to 40 times compared to OpenAI.
Companies built on OpenAI's infrastructure can transition to DeepSeek in minutes, making the switch nearly frictionless.
OpenAI's pricing for input tokens is $2.5 per million, while DeepSeek charges just $0.014, forcing major players to reconsider their pricing models.
European regulators are investigating whether DeepSeek copies OpenAIās data or censors politically sensitive topics.
While U.S. AI startups received nearly $100 billion in funding in 2024, Europe lagged with only $15.8 billion, highlighting the need for accessible alternatives.
The company claims its latest model was trained with just $6 million in computing power, far lower than its American rivals, though some analysts are skeptical.
While startups flock to DeepSeek, established players like Nokia and SAP prioritize security, compliance, and ecosystem integration over cost savings.
Forward Future Takeaways:
DeepSeekās aggressive pricing is forcing a shift in AI economics, pressuring major firms to reduce costs and innovate faster. While startups welcome the competition, concerns over security, transparency, and potential regulatory roadblocks remain. Whether DeepSeek becomes a long-term disruptor or a short-lived challenger will depend on its ability to navigate political and market challenges. ā Read the full article here.
š« FORWARD FUTURE UNIVERSITY
A Practical Guide to Choosing the Right AI Model
In the rapidly evolving landscape of artificial intelligence, OpenAI has recently introduced a powerful new reasoning model: ChatGPT o3 mini. As organizations and individuals increasingly rely on AI for various tasks, understanding the strengths and limitations of each new model becomes crucial for making informed decisions. Let's compare the new o3 mini to ChatGPT 4o and explore what makes each model unique and how to choose the right one for your needs.
The Rise of ChatGPT o3 mini: Speed and Efficiency in Focus
OpenAI's latest addition to their reasoning model family, o3 mini, represents a significant step forward in specialized AI capabilities. Designed to focus on reasoning, speed, and cost-effectiveness, this model brings impressive performance to technical tasks while maintaining efficiency. What sets o3 mini apart is its optimization for STEM-related problems and coding challenges, making it particularly valuable for developers and technical professionals.
The model introduces several noteworthy capabilities that distinguish it from its predecessors. ā Continue reading here.
šµļø OPERATOR REVIEW
OpenAIās Operator: A Glimpse Into the Future of AI AgentsāFlaws and All
The Recap: OpenAIās new AI agent, Operator, promises to automate everyday tasks like shopping and scheduling but struggles with erratic behavior and frequent hiccups. While it hints at a future where AI handles real-world tasks independently, its current form feels more like an insecure intern than a reliable assistantāan observation made by Kevin Roose of The New York Times.
Operator can browse the web and complete tasks like ordering from Amazon, booking appointments, and responding to messages with minimal human input.
It successfully ordered lunch on DoorDash, responded to LinkedIn messages, and even made money filling out online surveys.
However, it frequently required user intervention, often getting stuck in loops or asking excessive confirmations before executing tasks.
The AI is blocked from certain websites like The New York Times, Reddit, and YouTube, limiting its functionality.
It refused to engage in gambling and struggled with CAPTCHA tests, reinforcing existing safeguards against AI misuse.
Despite being marketed as a productivity tool, many tasks took longer to complete than if done manually.
The $200-a-month price tag makes it hard to justify, especially given its current limitations.
Forward Future Takeaways:
While Operator is more of a proof-of-concept than a must-have tool, it showcases the direction AI agents are headingātoward performing autonomous online tasks. Future iterations could become more powerful, potentially replacing some human workers, but they also raise concerns about AI-driven automation, cybersecurity risks, and the growing presence of bots on the internet. If left unchecked, we could be headed toward a āself-driving internetā where AI agents interact with each other more than with humans. ā Read the full article here.
š°ļø NEWS
Looking Forward
š©āš« Jensen Huang Says Get AI Tutor: NVIDIAās CEO believes AI tutors can enhance learning and work, not replace jobs. He urges everyone to start using AI educators now.
š° StackAdapt Raises $235M for AI Ads: The Toronto-based startup secured major funding to expand its AI-powered programmatic advertising. Backers include Ontario Teachersā Pension Plan and Intrepid Growth Partners.
š± Snap Unveils AI Image Model: Snap's new text-to-image AI runs on mobile devices, generating high-res images in 1.4 seconds. It will soon power AI Snaps and Bitmoji backgrounds.
š Make Perplexity Your Default Search: AI-powered Perplexity offers ad-free, research-friendly search results. Learn how to set it as your default in Chrome and Firefox for better queries.
šÆ Four Chinese AI Startups to Watch: Beyond DeepSeek, firms like Stepfun, ModelBest, Zhipu, and Infinigence AI are pushing AI innovation. They focus on AGI, efficiency, and infrastructure despite U.S. sanctions.
š½ļø VIDEO
The Tipping Point: OpenAI Unveils "Deep Research"
Matt explores OpenAIās Deep Research, an advanced AI agent that autonomously scours the web and generates fully cited research papers in minutes. Powered by OpenAIās 03 model, it excels in reasoning, source analysis, and expert-level synthesis. Get the full scoop! š
š„ FF INTEL
Got a Hot Tip or Burning Question?
Weāre all ears. Drop us a note, and weāll feature the best reader insights, questions, and scoops in future editions. Letās build this thing together.
šµ Hit the button below and spill the tea!
Reply