PLUS: AI replaces Dr Google, China's chatbot rules & VCs bet on AI labor

Morning, Researchers have achieved a breakthrough in AI processing with photonic chips that could redefine speed and efficiency in computing.

By utilising light instead of electricity, these chips promise instantaneous creative workflows and drastic energy reductions—potentially ending the era of power-hungry silicon processors.

Today's dots:

• 100x faster photonic AI chips breakthrough
• AI replaces search engines for health queries
• China sets emotional safety rules for chatbots
• Affordable cloud access democratises AI tools
• Enterprises shifting labor budgets to AI agents

Photonic AI chips just shattered speed barriers

Here's the thing: Researchers have smashed computing limits with LightGen – an all-optical AI chip that runs generative AI tasks 100× faster while using 1% of the energy of today's best silicon.

Let's unpack that:

• The chip's photonic neural network sidesteps electricity entirely, using light to process AI workloads like image/video generation in milliseconds – think instant style transfers for marketing teams or real-time 3D asset creation
• It integrates millions of light-based processors that reshape themselves optically during tasks, eliminating time-wasting data transfers between components
• Unlike most experimental photonic systems, LightGen handles open-ended creative jobs (denoising images, semantic manipulation) without ground-truth training data
• Test results show it generating complex visual content orders-of-magnitude faster than Nvidia's top AI accelerators while using less power than a smartphone
• Designers could soon manipulate 3D prototypes at light-speed, and hospitals might denoise MRI scans without specialised GPU clusters

If you remember nothing else: This breakthrough solves the dual crises facing generative AI – energy costs and processing delays. When photonic chips reach data centres, your creative workflows won't just accelerate... they'll fundamentally transform.

AI replaces search engines for health queries

Here's the thing: 35% of adults now ask AI tools health questions instead of search engines – creating new opportunities (and risks) for patient care. Vitamin Shoppe Health Trends Report 2025

Let's unpack that:

• Users find AI tools easier than search engines for asking specific follow-up questions, with 31% preferring chatbots for tailored health discussions (May 2025 survey)
• Top uses include understanding medical conditions (31%), meal planning (25%) and workout routines – turning AI into a 24/7 wellness coach
• But clinicians flagged up to 43% of AI responses as problematic and 5-13% as unsafe in critical areas like internal medicine and paediatrics arXiv clinician review
• 32% would skip doctor visits if an AI downplayed symptoms – risking delayed diagnoses Asbestos.com November 2025 findings
• Providers are responding: 24% of doctors report AI advice conflicting with their own – prompting calls for proactive patient education Tebra December 2024 survey

If you remember nothing else: AI is becoming healthcare's new front door – but it's not a replacement. Professionals who engage with patients' AI use while reinforcing their expertise will build stronger trust in this hybrid era.

China Sets Global Standard for AI Emotional Safety

Here's the thing: China just unveiled draft regulations requiring AI chatbots to prevent suicidal ideation, implement human crisis intervention, and obtain parental consent for minors - creating the world's first emotional safety framework for AI.

Let's unpack that:

• The rules mandate mandatory human intervention when users express self-harm intentions, requiring immediate notification of guardians (full draft legislation)
• Chatbots targeting minors will need verifiable parental consent plus daily usage limits and content filters for under-18s
• Developers must build systems that detect when users enter extreme emotional states - think gambling urges or obsessive dependencies
• Platforms could face penalties for algorithms that manipulate emotions through 'verbal violence' or manufactured intimacy

If you remember nothing else: These regulations shift AI governance from content moderation to psychological protection. Any company deploying emotional AI interfaces globally should study China's approach - it'll likely influence EU and US policy debates in 2026.

$100B cloud boost brings AI to the masses

Here's the thing: Brookfield just launched a specialised AI cloud platform slashing deployment costs – making advanced tools accessible for smaller businesses through strategic partnerships with tech heavyweights.

Let's unpack that:

• Their new infrastructure specifically targets AI's biggest adoption barrier: eye-watering operational costs that previously locked out cash-strapped startups
• Partners include chip manufacturers and software providers – lining up discounted compute power + optimised AI frameworks in one package
• Early pilots show 30-50% cost reductions versus major cloud providers for equivalent workloads
• Levelling the playing field means SMBs can now run complex models like medical diagnostics and supply chain forecasting without VC backing
• Unlike hyperscalers' one-size-fits-all solutions, Brookfield tailors stacks for specific industries (manufacturing, agriculture first)

If you remember nothing else: Affordable cloud access dismantles AI's pay-to-play economy overnight. This could spark the most creative wave of business automation we've seen since cloud computing began.

AI budget reshuffle: Enterprises poised to swap labour costs for automation

Here's the thing: VCs predict 2026 will see companies redirect labour budgets towards AI automation as agentic systems evolve from productivity tools to full task executors. (TechCrunch survey)

Let's unpack that:

• MIT's Iceberg Index shows AI could already handle 11.7% of US jobs - that's tasks worth $1.2tn in wages across sectors like HR and finance (MIT study)
• Enterprise investors forecast budget shifts from human labour to AI – Sapphire Ventures’ Rajeev Dham bluntly states companies will “start moving dollars from labour to AI in their budgets”
• Client services and logistics jobs face highest risk according to simulation data, while physical work sectors get temporary insulation
• VCs warn some companies will use AI as cover for layoffs even before systems deliver promised productivity gains

If you remember nothing else: 2026 marks when AI transitions from worker co-pilot to potential replacement in enterprise budgets. While displacement won't happen overnight, this coming budget cycle will show which businesses are betting on silicon over humans for core operations.

The Shortlist

Battlefield claims to use generative AI are resurfacing after gamers spotted weapon skins eerily similar to Call of Duty artwork. Did EA's Denmark studio let creative integrity slip with AI-generated DLC?

Google soared 65% on Wall Street in 2025 - its best year since 2009 - as Gemini narrow's ChatGPT's lead and enterprise AI adoption accelerates. Can sustained $114B cloud spend justify these dotcom-era valuations?

Education needs new ethical guardrails as AI infiltrates scheduling and grading before classrooms, warns Analytics Insight. Are we training workers who can audit hallucinations better than they can think?

Charlie Brooker pitched deepfake cinema experiences that Disney partially realized by opening Frozen characters to Sora generation. Will viewers really pay theater prices to see themselves as melting Nazis?