AI – Tech DC

The Future in a Chart: Understanding AI Risk Before It’s Too Late

October 14, 2025 / AI

We’re living through a technological revolution that could end in utopia, normalcy, or extinction. If that sounds dramatic, consider this striking chart from the Dallas Federal Reserve that maps three wildly different futures based on how AI develops. One line shows modest GDP gains. Another rockets toward post-scarcity abundance. The third plummets to zero—representing human extinction from misaligned AI. This isn’t science fiction. It’s economic modeling from one of America’s most conservative institutions.

The Godfather of AI Issues a Warning

In this essential interview with Jon Stewart, Geoffrey Hinton—the “Godfather of AI” who won the 2024 Nobel Prize in Physics for pioneering neural networks—explains why even their creators don’t fully understand these systems. When Stewart asks if AI systems can make decisions based on their own experiences, Hinton replies: “In the same sense as people do, yes.” Hinton believes that AI will develop consciousness and self-awareness, and when asked if humans will become the second most intelligent beings on the planet, responds simply: “Yeah.” This isn’t speculation from a fringe theorist. This is the man whose work made modern AI possible, and he left Google specifically to warn about what he helped create.

Why the Engineers Are In Over Their Heads

If Anybody Builds It, Everybody Dies: Why Superhuman AI Would Kill Us by Eliezer Yudkowsky and Nate Soares exposes a disturbing paradox: the people who best understand AI’s existential risks—including OpenAI’s CEO, Google DeepMind executives, and Turing Award winners—are the same ones accelerating toward them in what the book calls a “suicide race.” The engineers building these systems are brilliant and well-meaning, but they can’t explain what happens inside large language models or control what these systems ultimately want. Yudkowsky himself inspired OpenAI’s founding, a move his own institute now views as having “backfired spectacularly.” The book’s core warning: by the time the danger becomes obvious to everyone, it will be too late.

Policymakers Are Starting to Ask Questions

While researchers warn about existential risk, Senator Bernie Sanders is addressing AI’s immediate economic impact, warning that nearly 100 million U.S. jobs could vanish within a decade. His Senate report proposes solutions including a “robot tax” and reduced work weeks. Whether you agree with specific proposals or not, the fact that policymakers are finally engaging with these questions matters—because the people building AI are telling us they don’t have all the answers.

The Dallas Fed chart shows we’re not choosing between similar futures—we’re choosing between radically different outcomes for civilization. Watch the Hinton interview. Read the book. And demand that the conversation about AI’s future includes more voices than just those racing to build it. The window to get this right is closing.

AI Goes Local: Run Powerful Models Right on Your Computer

August 6, 2025 / AI, App, Apple, Computer Hardware

The AI revolution just got more personal. While most of us have grown accustomed to typing prompts into ChatGPT or Claude through our browsers, a new wave of tools is bringing that same AI power directly to your desktop—no internet required, no monthly subscriptions, and complete privacy.

Two standout applications are making local AI accessible to regular users: Ollama for text generation and DiffusionBee for creating images. Both represent a significant shift toward democratizing AI technology.

Ollama: Your Personal ChatGPT

Ollama recently launched a partnership with OpenAI and released a new GUI application for macOS and Windows, transforming what was once a command-line tool into something anyone can use. The app lets you download and chat with various AI models, including OpenAI’s new open-source gpt-oss models available in 20B and 120B parameter sizes.

This is significantly more user friendly than it used to be. The new interface supports file drag-and-drop, letting you upload PDFs or documents and have conversations about their contents. It even includes built-in web search capabilities, though you’ll need to enable that feature.

The performance trade-off is real but workable. The smaller 20B model can run on systems with just 16GB of memory, though more RAM certainly helps. On my M4 Mac with 32GB, responses often have a delay of a minute or so, but then come faster than I can read them. This is noticeably slower than the cloud-based services we’re used to. The key difference is that everything happens on your machine, with complete privacy.

DiffusionBee: AI Art Without the Cloud

For image generation, DiffusionBee brings Stable Diffusion to your desktop with remarkable ease. This completely free application runs locally on your computer with a one-click installer and no technical knowledge required.

The app includes text-to-image generation, image-to-image transformation, inpainting, and upscaling features—essentially everything you’d find in premium online services. Your prompts, models, and generated images never leave your device, addressing privacy concerns that many professionals in the DC area have about cloud-based AI tools.

The quality won’t match the latest offerings from Midjourney or DALL-E, but for many use cases—creating presentation graphics, brainstorming visual concepts, or just experimenting with AI art—it’s more than sufficient.

Why This Matters Now

This shift toward local AI addresses several concerns that have limited adoption in professional settings. Government contractors, lawyers, and consultants who handle sensitive information can now experiment with AI without worrying about data leaving their systems. Small businesses can access powerful AI capabilities without ongoing subscription costs.

The models available today are roughly equivalent to GPT-3.5 or early GPT-4 performance levels—a step behind the cutting edge, but still remarkably capable for most tasks. And unlike cloud services, there are no usage limits, rate limits, or monthly bills.

Both applications install in minutes and start working immediately. Ollama downloads models as needed (expect several gigabytes per model), while DiffusionBee includes everything required to start generating images.

The local AI movement represents more than just an alternative to cloud services—it’s about putting powerful technology directly into users’ hands. For the DC community, where data privacy and independence matter, these tools offer a compelling glimpse of AI’s more distributed future.

Download Ollama at ollama.com and DiffusionBee at diffusionbee.com. Both are free and work best on modern computers with at least 16GB of RAM.

AR Glasses Are Coming—But Will Anyone Really Wear Them?

June 24, 2025 / AI, Apple, Electronics

After years of hype, AR glasses are finally on the horizon. Snap says its next-gen smart glasses will launch in 2026. Meta has already sold over two million of its Ray-Ban smartglasses. And it’s reportedly working on a new version instead of next updating its Quest line of VR headsets. Apple is rumored to be quietly working on its own version. These will be much less obtrusive than big VR headsets—they look (mostly) like regular glasses and promise real-time translation, video calls, and turn-by-turn directions, all floating in front of your eyes.

It sounds amazing. But will people actually wear them?

We’ve already seen how virtual reality (VR) goggles, while fun for gaming, are not yet regularly used by most people. Typically, people are wowed by the technology, but don’t want to strap something to their face that cuts them off from the real world. Think of it like an amusement park. It’s really great and fun, but you wouldn’t want to go every day or every week.

I’ve used both the Apple Vision Pro and the Meta Quest 3 VR headsets. Both are fun, but even if they were lighter and cheaper, they’re a niche product because they’re a big thing strapped to your head. AR is much more subtle—but it still involves wearing something on your head.

For AR glasses to go mainstream, they need to solve real problems—without being awkward or invasive. Think language translation while traveling, guided workouts, hands-free help for delivery drivers, or facial recognition to remind you who that work associate is. It could be an easy way to interact with ever more ubiquitous AI tools. But wearing something on your face is much more difficult and personal than wearing a watch. If the tech feels weird or doesn’t match people’s personal tastes, it won’t be widely used.

For now, the promise is exciting. The reality? Still very much TBD.

AI’s Future: Utopia or Nightmare

June 12, 2025 / AI, Software

Artificial intelligence is racing ahead—so fast that some experts say we’re living through the most important tech shift since electricity. But where’s it all going?

Right now, there are two main visions of the future:

The Optimists – Leaders like Sam Altman (OpenAI) and Dario Amodei (Anthropic) see AI as a game-changer for productivity, healthcare, and science. They predict tools that will help us work faster, solve complex problems, and maybe even extend human lifespans.
The Pessimists – Others, like Geoffrey Hinton (the “Godfather of AI”) and Yann LeCun (Meta), warn we might be going too fast given that we don’t fully understand what we’ve created and how it works. Hinton even gives a 10–20% chance that advanced AI could pose an existential risk.

What’s on the Line?

Jobs: Most expect some degree of job losses. Dario Amodei, CEO of Anthropic has warned that AI technology could eliminate half of all entry-level white-collar positions within the next five years. The prediction suggests unemployment could reach 10-20% as AI systems transition from augmenting human work to replacing it entirely.
The Economy: AI might spark massive productivity growth for companies. But who benefits? That’s why people are now seriously discussing things like Universal Basic Income (UBI) to help distribute gains fairly.
Global Competition: The U.S. and China are locked in an AI arms race. Some experts are calling for international agreements—like we have for nuclear weapons—to slow things down and focus on safety.
Daily Life: AI is becoming your assistant, tutor, creative partner, and even your co-worker. But there are risks too: misinformation, deepfakes, a reliance on machines to think for us, and humans becoming more disconnected from each other.

So What Now?

Even the most bullish tech leaders agree: we need to be thoughtful. That means building systems we can trust, creating smart policy, and helping people adapt.

At TechDC, I help individuals and small businesses learn how to actually use AI in real life. If you’re curious, reach out.

AI