🍭 LLMs, Brains, And Fake Intelligence

Good morning. My gaming PC just found out its RAM is too poor for Black Friday and too cheap for OpenAI, so it’s sitting in the corner like a rejected organ donor.

Let’s dig in. 👇

🍭 What’s Cookin’:

• Why LLMs still can’t “think” like you think they can

• How AI data centers just hijacked Black Friday RAM

• Amazon, Google & OpenAI, and States vs. Feds

• ToolBox™

_ToolBox™
🧰 5 BRAND NEW AI LAUNCHES

🧪 Cracked.ai
→ Spin up AI “influencer” agents that post, comment, and reply across social like a squad of tireless creator clones handling your distribution for you.

🧠 Waylight
→ An on-device AI assistant that remembers your tabs, meetings, docs, and messages so you can ask “what did I agree to yesterday?” instead of hunting through 40 windows.

🎯 Seefy
→ AI event intel that scans the global conference universe, flags the shows that actually drive pipeline, and helps you capture target accounts before anyone hits the expo floor.

📚 Unfold
→ Generate interactive, animated courses on any topic with Gemini 3, tuned to how you learn and what you actually want to master.

🖥️ Ripplica
→ Record a browser workflow once and let AI agents replay it forever, automating web tasks without touching a single API or line of code.

“Thinking”???
🤯 LLMs Will Never Be Intelligent? Cool, That Wasn’t the Job

The Bite: A new essay says LLMs will never be truly intelligent because language and thought are different things.

Fair, in a neuroscience sense. Your brain can plan, plot, and panic without narrating it.

But if you actually use AI, the better question isn’t “Is this real intelligence?” It’s “Is this fake-looking intelligence useful enough to reshape work anyway?”

Especially when it still whiffs on long-term planning and emotional nuance.

Snacks:

• The core claim: LLMs remix language, they don’t actually think.

• The science bit: Brain language areas ≠ reasoning or planning systems.

• The ceiling: Scaling text + GPUs won’t magically unlock “digital minds.”

• The marketing: AGI talk on top of this is mostly fundraising theater.

• The reality check: You still feel the gaps in long tasks and human emotion.

Why it Bites: If you track AI news, you’re stuck between “superintelligence is imminent” and “it’s just autocomplete.”

This argument is the useful middle: stop judging models by sci-fi branding and leaderboard crumbs.

For AI generalists, the new filter is simple: does this release actually improve long-run workflows, goal tracking, and human-facing nuance?

If the answer is “no, but the demo looks cool,” you can safely scroll past the announcement.

Steal This Prompt
🎨 UI Generator: Your Website If Apple Designed It

Turn any website or product into a slick “designed by Apple” concept mockup in a few seconds, without learning Figma or channeling Jony Ive in the mirror.

Use it to reimagine your SaaS dashboard as a keynote slide, redesign your local dentist’s site like a billion-dollar wellness startup, or turn your personal blog into something that looks suspiciously VC-backed.

Workflow:

1. Click this link (Prompt)

2. Paste into GPT

3. Replace the #’s with your site, product, or page details

4. Watch it cook you a painfully clean, minimal Apple-core redesign

🍭 Try the prompt on Snack Prompt 🍭

Token Factory
🤯 Launch of Nebius Token Factory — AI at production scale

We’re announcing Nebius Token Factory, the next evolution of Nebius AI Studio:

A platform built for production-scale AI.

Token Factory powers dedicated inference, fine-tuning, and governed collaboration for open-source models like Llama, Qwen, DeepSeek, and GPT-OSS — all with enterprise-grade reliability, compliance, and full visibility.

_{Everything Else}
🧠 You Need to Know

📈 OpenAI, Google say AI is reshaping go-to-market

→ Leaders from OpenAI, Google Cloud and GTMfund say AI is letting startups move faster and personalize outreach, but still requires classic marketing skills and domain expertise.

🧱 Supabase CEO on building a $5B database company

→ Supabase CEO Paul Copplestone explains how sticking to product vision, investing in Postgres and turning down some lucrative enterprise deals helped the company grow to a $5B valuation.

⚖️ US faces federal vs state clash on AI rules

→ As Congress inches toward national AI standards, tech firms push for federal preemption while lawmakers debate how far to limit state-level AI safety laws without weakening consumer protections.

🚨 Developers highlight gender and race bias in AI tools

→ A Perplexity user’s experience with sexist responses prompts experts to note that large models can still encode bias, and that asking systems if they are discriminatory does not reliably reveal those patterns.

📨 Amazon workers warn Jassy over AI, climate and jobs

→ More than 1,000 Amazon employees signed an open letter urging CEO Andy Jassy to address the climate, water and labor impacts of the company’s AI expansion and to include workers in AI-related decisions.

_{Consumers VS AI Infra}
👾 AI Ate Your RAM Deals

The Bite: If you went hunting for Black Friday RAM deals this year and thought, “Wait… these all suck,” you’re not crazy.

DDR5 kits that were ~$150 a few months ago are now “on sale” for $350–$400 with fake markdowns and bonus junk thrown in.

The culprit isn’t gamers or crypto this time.

It’s AI data centers inhaling the same memory you need for your rig.

And they’re not just nudging RAM prices; they’re also pushing up power demand that eventually shows up on your electricity bill.

Snacks:

• Black Friday is cap: Most “deals” are higher than spring prices.

• AI eats first: SK Hynix, Samsung and friends are prioritizing server RAM for AI giants.

• Consumer scraps: Even DDR4 prices are drifting up — when you can find stock.

• Price whiplash: CyberPowerPC says RAM is up 500%, SSDs 100%, and they’re raising PC prices.

• Power tax: Data centers for AI are set to more than double electricity use by 2030, and grids are already pricing that in.

Why it bites: PC bros are getting the bill for the AI boom they read about in press releases.

Every “we’re investing billions in AI infrastructure” headline quietly translates into tighter memory supply, higher component prices, and more strain on power grids that households ultimately help pay for.

This is the new normal: when OpenAI, Anthropic, or whoever signs another monster GPU + HBM deal, it doesn’t just move their valuatio;
It ripples down into what you pay to upgrade your RAM, your SSD, and eventually your monthly energy costs.

Today it’s a bad year to build a gaming PC. Tomorrow it’s everything with a plug.