Good morning. A pro-AI subreddit has banned over 100 users who believe they've created AI gods or become gods themselves through ChatGPT conversations. Apparently even the most enthusiastic AI boosters have their limits when the chatbots start convincing people they're deities.

— The Deep View Crew

In today’s newsletter:

• 🥦 AI for Good: Predicting child malnutrition

• ⚡️ OpenAI roadmap envisions ChatGPT as “Super-Assistant” by 2025 

• 🏛️ Musk's $5 billion AI bet marks shift from politics to tech

🥦 AI for Good: Predicting child malnutrition

What happened: Researchers at USC, Microsoft's AI for Good Lab, Amref Health Africa and Kenya's Ministry of Health built a machine learning model that can predict acute child malnutrition in Kenya up to six months before symptoms appear.

• The team trained the AI on clinical data from more than 17,000 Kenyan clinics, combined with satellite data tracking crop conditions and rainfall patterns. The system analyzes these inputs to identify patterns that precede spikes in severe malnutrition.

• In testing, the model achieved 86% predictive accuracy for six-month forecasts and 89% accuracy for one-month predictions, significantly outperforming traditional forecasting methods.

Why it matters: Kenya faces a malnutrition crisis, with about 5% of children under five (roughly 350,000 kids) suffering acute malnutrition. In some arid regions, that number reaches 25%. Until now, officials relied on expert judgment and historical trends to predict food crises—tools that struggle to anticipate sudden shifts.

"The best way to predict the future is to create it using available data for better planning and pre-positioning," said Murage Kiongo of Kenya's Ministry of Health. The early warning system gives authorities and NGOs up to six months of lead time to pre-position food supplies and deploy health workers before malnutrition rates soar.

The framework could extend beyond Kenya—over 125 countries use the same health information system, meaning similar AI tools could be deployed across many low- and middle-income nations.

Together AI: The AI Acceleration Cloud for Inference & Turbocharged GPU Clusters

Unify every stage of your generative AI workflow - launch, customize, and scale on one platform built atop NVIDIA Blackwell HGX B200 and GB200 NVL72 GPUs.

What You Get with Together AI

• Serverless & Dedicated Inference

  Launch endpoints in minutes - fully SOC 2 and HIPAA compliant, with AWS deployment options for ironclad security.

• 200 + Open-Source Models

  Tap into a library of over 200 high-performance models - DeepSeek-R1-0528, Llama, Qwen, Flux - via our OpenAI-compatible API.

• Full & LoRA Fine-Tuning

  Own your model: from full-parameter fine-tuning to lightweight LoRA adapters - now with preference optimization for sharper results.

• Custom GPU Clusters

  Scale from 16 to 1,000+ GPUs (Grace Blackwell GB200 NVL72, HGX B200, H200, H100) - engineered for throughput with InfiniBand, NVLink, and Together Kernel Collection. Includes 99.9% SLA and expert AI advisory.

• Transparent, Cost-Smart Pricing

  Utility pricing per million tokens I/O, and GPU clusters starting at $1.75/hour - no surprises, no lock-in.

Get Started Now

Start your free inference trial and, when you’re ready for hyperscale training, our team will design and reserve the perfect GPU cluster for your roadmap.

⚡️ OpenAI roadmap envisions ChatGPT as “Super-Assistant” by 2025 

Newly leaked internal strategy documents show OpenAI’s plans to transform ChatGPT from a simple chatbot into a versatile “super-assistant” in the coming year.. The confidential “ChatGPT: H1 2025 Strategy” paper, revealed through a DOJ antitrust trial (where Google will have to outlay $500m), outlines an AI assistant that deeply understands individual users and can plan vacations, make restaurant bookings, manage emails and even execute tasks across apps or the web on a user’s behalf. OpenAI describes this “T-shaped” super-assistant as having broad everyday skills. It’s envisioned as “your interface to the internet,” available via ChatGPT’s app, phone, or even integrated into voice assistants.

According to the leaked documents, OpenAI is using a five-tier scale to measure AI capabilities. Current ChatGPT models are considered early-stage, focused on conversation and reasoning, while the next milestone – Level 3 “agents” – denotes AI systems that can take autonomous actions. Level 5, the ultimate stage, is described as AI that could handle the work of an entire organization – essentially a form of superintelligence. OpenAI’s roadmap positions ChatGPT’s evolution around Level 3 by mid-2025, moving beyond text responses to actually completing multi-step tasks for users.

Product and monetization ambitions also emerge in the leaks. OpenAI plans to begin the ChatGPT super-assistant rollout in the first half of 2025, then leverage its expanded capabilities for revenue in the latter half of the year. “We will build a super-assistant that can generate enough monetizable demand to pursue these new models in H2,” the document states. In practice, that likely means driving usage with advanced assistant features – potentially leading up to new subscription offerings or more powerful model releases by late 2025. The strategy underscores that while ChatGPT’s user base is surging, future growth (and profit) will depend on turning the chatbot into an indispensable, do-it-all AI assistant.

✂️ Cut your QA cycles down from hours to minutes

If slow QA processes bottleneck you or your software engineering team and you’re releasing slower because of it — you need to check out QA Wolf.

QA Wolf's AI-native service supports both web and mobile apps, delivering 80% automated test coverage in weeks and helping teams ship 5x faster by reducing QA cycles to minutes.

With QA Wolf, you get:

✅ Unlimited parallel test runs

✅ 15-min QA cycles

✅ 24-hour maintenance and on-demand test creation

✅ Zero-flake guarantee

The result? Drata’s team of 80+ engineers saw 4x more test cases and 86% faster QA cycles.

No flakes, no delays, just better QA — that’s QA Wolf.

Schedule a demo to learn more

• Media has a blind spot when covering the AI panic

• Salesforce buys Moonhub, a startup building AI tools for hiring

• Netflix co-founder Reed Hastings joins Anthropic’s board of directors

• McKinsey is using AI to create PowerPoints and take over junior employee tasks

• Digg’s founders explain how they’re building a site for humans in the AI era

• China is gaining ground in the global race to develop AI agents

• AI cheating is so out of hand in America’s schools that the blue books are coming back

• Writer: Product & Technical Enablement Manager

• Figure: Global Supply Manager - Mechanical

• Scale AI: Applied AI Engineer, Autonomous Agents

• Circuit Tracer: Anthropic's open source tool to see how AI thinks

• ElevenLabs Conversational AI 2.0: New features to create better voice agents

🏛️ Musk's $5 billion AI bet marks shift from politics to tech

Elon Musk is turning back to his businesses with an ambitious new financing spree. The billionaire is raising a $5 billion debt package for his artificial intelligence startup xAI, led by Morgan Stanley as underwriter. The debt deal carries a double-digit interest rate and includes multiple tranches.

Here's the breakdown: Early commitments have already topped $3.5 billion ahead of a June 17 deadline. In parallel, Musk is seeking about $650 million in fresh funding for Neuralink, his brain-computer interface venture.

• The funds will bankroll xAI's expansion of AI infrastructure, including a massive new data center in Memphis, Tennessee.

• Neuralink's latest round comes as the company's implant enters human trials in three countries, with five patients already implanted.

These moves come days after Musk's exit from an official government role in Washington. Late last month, his 130-day term as head of the "Department of Government Efficiency" in Donald Trump's administration expired, though he will continue as an unofficial adviser and Trump said "Elon is really not leaving" from his orbit.

Musk's Washington stint had drawn backlash after he spent roughly $300 million backing Trump. Tesla customers soured on the brand, and facing investor pressure amid sliding Tesla sales, Musk announced he would step away from politics and return to "spending 24/7 at work."

What is xAI?

xAI is Musk's answer to OpenAI and Google's DeepMind — an AI startup he founded in 2023 with the mission to "understand the true nature of the universe." Musk has said xAI's goal is to build a "maximally curious" AI that seeks deeper truth rather than being constrained by programmed moral rules.

• The venture assembled a team of top machine-learning researchers from OpenAI, Google and other tech giants.

• The new $5 billion war chest is intended to fund advanced AI model training infrastructure and talent as xAI races to catch up with industry leaders.

xAI is also launching a $300 million share sale that values the combined group at $113 billion, validating the March takeover pricing when xAI acquired X. The deal will allow staff to sell shares to new investors and is expected to be followed by a larger investment round.

The competitive landscape is fierce. OpenAI enjoys billions in backing from Microsoft. Google DeepMind is drawing on Google's vast resources. Rival startup Anthropic recently secured a $3.5 billion investment round at a $61.5 billion valuation, boosting development of its Claude AI assistant.

But xAI is gaining distribution partnerships: Microsoft announced it was making xAI models available to its cloud computing customers, while messaging app Telegram agreed to distribute Grok to its 1 billion users.

Zoom in: Neuralink, founded by Musk in 2016, is developing implantable brain chips to connect human minds directly to computers. The technology could help restore movement and communication to people with paralysis, and ultimately enable "brain-machine symbiosis."

This year Neuralink reached a milestone: its first human trials received regulatory approval, and five patients have been implanted with the device, allowing them to control digital devices with just their thoughts. The new $650 million funding round values the company at $9 billion, up from $3.5 billion in 2023, and will accelerate research and clinical programs.

But Neuralink faces growing competition. Last month, rival startup Paradromics completed its first human brain implant test at the University of Michigan. The company's Connexus Brain-Computer Interface was safely implanted and removed during epilepsy surgery in about 20 minutes. Other competitors include Synchron, backed by Jeff Bezos and Bill Gates, and Precision Neuroscience. The field has attracted more than $2 billion in investment as companies race to bring brain-computer interfaces to market.

Musk's reorientation could reverberate across the tech landscape. In the AI domain, xAI's massive funding boost positions it as a new heavyweight contender. Access to $5 billion gives xAI the means to purchase enormous computing power and attract top-tier AI talent.

Musk's dual role as owner of X and head of xAI also hints at synergies. xAI could leverage X's vast real-time social data to train its models, potentially giving it an edge in conversational AI.

For Neuralink, success would validate Musk's vision of merging man and machine. It aligns with his broader concern about AI: he has often argued that improving human cognition through implants might be necessary if artificial superintelligence arrives within this decade.

By publicly distancing himself from partisan battles, Musk is attempting to rebuild goodwill with customers, advertisers and regulators who grew wary of his polarizing political forays. His message is unmistakable: he's "super focused" on X, xAI, Tesla and SpaceX going forward. Whether juggling four frontier technology companies while maintaining informal political influence proves sustainable remains to be seen. But the $5.65 billion funding blitz sends a clear signal that Musk is betting his future, and his fortune, on technology disruption, not political theater.

🤔 Your thought process:

Selected Image 1 (Left):

• “That one random black brick on the chimney where the chimney sits on the roof is not the sort of thing an AI would add in.”

• “Tough one. The sloppy mortar cries out that it was done by a human. No details are really visible in the AI version- no flashing for instance.”

Selected Image 2 (Right):

• “The joins in [the other image] look so shoddy I didn't think they could be real. But apparently there are truly incompetent builders out there.”

• “I have never seen brickwork like [the other image]”

💭 Poll Results

Here’s your view on ”Do you think we'll see a 50% wipe of entry-level jobs in next 5 years”?

Yes (43%):

• “In just a few months, AI has advanced leaps and bounds. Five years is plenty of time learn what it needs to expertly perform the tasks of an entry-level job. Companies will likely be eager to save where they can.”

• “People generally underestimate the pace of technological progress. In this case, the very progress that is expected is being driven by the replacement technology itself, so the expected growth rate will be even greater.”

• “I believe we are seeing it some now, more in the next three years, so yes. In five years, it is possible for the human workforce to see a 50% wipe with AI doing many entry level jobs.”

No (47%):

• “I believe that those entering the workforce will need to understand AI and be able to utilize it in their jobs. It will make them much more productive, and not to mention, many of the younger generation already have been using AI and are relatively accustomed to it.”

• “I think the people looking for those jobs are not going to allow their futures to be taken over, so even though it might suit companies economically to replace humans with AI, the humans will rebel in my view - unless someone pays them to sit home and do nothing.”

Other (10%)

Thanks for reading today’s edition of The Deep View!

We’ll see you in the next one.

P.S. Enjoyed reading? Take The Deep View with you on the go! We’ve got exclusive, in-depth interviews for you on The Deep View: Conversations podcast every Tuesday morning. Subscribe here!

If you want to get in front of an audience of 450,000+ developers, business leaders and tech enthusiasts, get in touch with us here.