OpenAI to Charge 20k Per Month for PhD Level AI Agents?!

Also, Axios explores what AI leaders mean by “human in the loop”

Author

Jack Lajoie
March 06, 2025 • read time ~ 8 minutes

⚡️ Headlines

🤖 AI

AI Will Soon Dominate Ad Buying, Whether Marketers Like It or Not - AI tools are automating digital advertising, optimizing placements and targeting, but concerns arise over transparency and control. [The Wall Street Journal].

Google launches Gemini 2.0, its latest AI model that can 'think multiple steps ahead' - Google introduces Gemini 2.0, an advanced AI capable of autonomous problem-solving and planning, enhancing user assistance and web navigation. [Business Insider].

Users report emotional bonds with startlingly realistic AI voice demo - Users form emotional connections with a new AI voice demo, highlighting the technology's increasing realism and potential social implications. [Ars Technica].

Anthropic quietly removes Biden-era AI policy commitments from its website - Anthropic has discreetly eliminated references to previous AI policy commitments made during the Biden administration from its website. [TechCrunch].

Introducing Thunderforge: AI for American Defense - Scale AI leads Thunderforge, a Department of Defense initiative to integrate AI into military decision-making, enhancing operational efficiency under human oversight. [Scale AI].

🦾 Emerging Tech

Bored Ape Creator Yuga Labs Says SEC Closing Investigation in 'Huge Win' for NFT Sector - Yuga Labs announces the SEC is closing its investigation, marking a significant positive development for the NFT industry. [Decrypt].

Coinbase Eyes Tokenized Securities as Crypto Regulations Shift Under Trump Administration - Coinbase explores offering tokenized securities in response to evolving cryptocurrency regulations under the Trump administration. [The Block].

🤳 Social Media

YouTube Premium Lite: Ad-Free Viewing for $7.99/Month - YouTube launches 'Premium Lite' in the US, offering ad-free viewing for $7.99/month, excluding offline downloads and background play. [YouTube Blog].

🔬 Research

Alibaba shares surge after it unveils reasoning model - Alibaba's shares rose over 8% following the release of its new AI reasoning model, QwQ-32B, which matches the performance of larger models with fewer parameters. [Reuters].

Latest Turing Award winners again warn of AI dangers - Turing Award recipients Andrew Barto and Richard Sutton caution against rapid AI deployment without rigorous safety testing, emphasizing potential risks. [The Verge].

⚖ Legal

Trump wants Republicans to 'get rid of' the CHIPS Act. They may just ignore him. - President Trump urges Congress to repeal the $52 billion CHIPS Act aimed at boosting U.S. semiconductor manufacturing, but GOP lawmakers show reluctance. [Business Insider].

A judge blocked Musk's bid to stop OpenAI's for-profit conversion — but he'll still get his day in court - A judge denies Elon Musk's attempt to halt OpenAI's transition to a for-profit entity; the case proceeds to trial. [Business Insider].

🎱 Random

The LA Times' AI 'bias meter' looks like a bid to please Donald Trump - The LA Times introduces an AI 'bias meter' to evaluate opinion pieces, perceived as an effort to align with pro-Trump views, sparking criticism. [The Guardian].

🔌 Plug-Into-This

OpenAI is set to introduce specialized AI "agents" tailored for various professional tasks, with monthly fees reaching up to $20,000. These agents aim to enhance efficiency in fields such as sales, software development, and research.

  • Diverse Pricing Tiers: The AI agents will be offered at different price points: $2,000 per month for high-income knowledge workers, $10,000 for software developers, and $20,000 for agents capable of PhD-level research.

  • SoftBank's Investment: OpenAI investor SoftBank has committed to investing $3 billion in these agent products throughout 2025, underscoring the anticipated value and impact of these AI solutions.

  • Operational Costs: OpenAI's computing expenses are projected to reach $37.5 billion annually by 2029, highlighting the substantial resources required to support advanced AI operations.

🤖 OpenAI’s pricing strategy positions its AI agents pretty much exclusively as premium enterprise tools. But highly specialized high-cost automation for industries willing to pay for cutting-edge efficiency could be worth the investment for businesses. Suddenly, a Bloomberg terminal (which gives real-time data insights to financial professionals and costs $20,000 per year) doesn’t seem so expensive!

The concept of "humans in the loop" in AI systems addresses the integration of human oversight and intervention in artificial intelligence workflows to mitigate potential risks such as errors, job displacement, and societal disruption. The necessity and extent of human involvement remain dynamic as AI capabilities advance, making it increasingly challenging to determine optimal balance points between human and machine intelligence.

  • AI Assisting Humans: Current AI systems often require human prompts and instructions to function effectively, highlighting the collaborative nature of human-AI interactions.

  • Critical Control Transfers: Advanced AI agents may necessitate human intervention at key decision points to ensure appropriate outcomes, especially as their autonomy increases.

  • Human Oversight: The accuracy and reliability of AI outputs often depend on human review, underscoring the importance of human oversight in AI applications.

  • Evolving Skill Requirements: As AI systems become more integrated into various sectors, human workers will need to develop skills in overseeing AI processes and effectively delegating tasks to machines.

🧠 Implementation of “humans in the loop” can take various forms: AI assisting humans by following prompts and instructions, AI handing over control to humans at critical points, and humans reviewing AI outputs to ensure accuracy. Despite advancements, AI systems still generally lack the nuanced understanding inherent to human judgment, making human oversight crucial in complex or ambiguous situations (for now).

Google has launched an experimental "AI Mode" for its search engine, replacing traditional search results with comprehensive AI-generated summaries. This feature is currently available to Google One AI Premium subscribers.

  • AI Mode Overview: In AI Mode, users receive detailed AI-generated summaries with hyperlinks to cited webpages, eliminating the classic list of blue links. A search bar is provided for follow-up questions.

  • Subscription Details: Access to AI Mode requires a Google One AI Premium subscription, priced at $19.99 per month, which also includes additional cloud storage and exclusive AI features.

  • Gemini 2.0 Integration: The AI Mode is powered by a custom version of Google's Gemini 2.0 model, enhancing its ability to handle complex queries through advanced reasoning capabilities.

Google, facing competition from new alternatives like AI-led search options like ChatGPT and Perplexity, is likely eager to display equivalent experiences for users. It’s not exactly clear how much of traditional search will (or can) be replaced by AI search features, but Google will stake it’s own claim nonetheless to maintain dominance in the industry it helped pioneer.

 🆕 Updates

📽️ Daily Demo

🗣️ Discourse