I'm an Old Fart and AI Makes Me Sad
Veteran programmers reacting to an essay titled “I’m an Old Fart and AI Makes Me Sad” weigh excitement about AI’s capabilities against unease over its opacity, centralization, and impact on craft and careers. Many describe feeling disempowered by black‑box models controlled by a few corporations, worried about enshittification, job displacement, and a flood of shallow “good enough” content, especially in art and coding. Others counter that the underlying math is accessible, open-source models and local tools are improving quickly, and that AI can be a powerful new layer of abstraction and creativity—if society manages its incentives and use responsibly.
Empowerment vs. Threat to Work
- Many see AI tools as empowering: non‑programmers are learning to script, juniors resolve questions faster, and coders use LLMs for boilerplate, glue code, and learning new tech.
- Others feel AI competes directly with their hard‑won skills, devalues expertise, and accelerates “good enough” mediocrity. Past automation (e.g., ATMs) is cited as precedent for job displacement.
- There’s tension between work as meaning vs. work as burden; some would gladly offload drudge work, others feel their craft is being hollowed out.
Opacity, Understanding, and Control
- A core sadness: AI feels like an opaque black box of billions of weights, unlike traditional systems where you can follow logic line by line.
- Some argue this is just another abstraction layer (like CPUs/compilers) and the underlying math (gradients, matrices) is accessible; others stress that even with math and code, individual model decisions remain unintelligible.
- This loss of inspectability reduces a sense of agency and makes debugging, guarantees, and trust feel qualitatively different.
Openness, Resources, and Centralization
- Many worry that meaningful AI work requires massive proprietary datasets, compute, and corporate capital, concentrating power and “AI wealth.”
- Others counter that open models (LLaMA‑family, Mistral, Stable Diffusion, etc.) and local tooling (llama.cpp, Ollama) have rapidly improved and run on consumer GPUs, echoing the PC revolution.
- There’s concern about API lock‑in: if a major provider changes pricing or features, dependent products can vanish overnight.
Art, Media, and Cultural Impact
- Several commenters are deeply depressed by AI‑generated imagery and text: erosion of human intent, flooding of low‑effort “content,” difficulty trusting what’s real, and weakened incentives to engage with art as expression.
- Others care mainly about the end result: if an image or text is moving or useful, the process or authorship matters less.
- Analogies to photography vs. painting are frequent; some think AI is just another tool that will enable new art forms, others see it as fundamentally shallow collage.
Hype, Research Quality, and Future Trajectory
- Skeptics compare generative AI to crypto/metaverse hype: “empty calories” research, junk papers, grant‑chasing, and corporate ad‑tech focus.
- Supporters argue we’re early in a genuine paradigm shift; open research and local experimentation are vibrant, and capabilities are improving quickly.
- Many feel overwhelmed by the volume and pace of new papers and techniques, worrying about keeping skills current.
Coping Strategies and Attitudes
- Suggested responses: learn the basics of ML math, follow practical courses, tinker with local models, treat AI as another powerful but imperfect tool.
- Emotional reactions diverge: some are re‑energized and “having fun again”; others lean toward Luddite instincts, retreat to non‑digital pursuits, or accept that not every new tech must be personally mastered.