We are not going anywhere
gist.github.com
August 24, 2026
1 min read
44/100
Summary
A forecast predicts that AI systems will perform most software development because their output will be commercially acceptable at far lower cost, even when it falls short of the quality expected from human-led engineering. It expects businesses to accept software with “99.99” quality rather than “99.999” quality when the cost difference is substantial, and predicts that consumer expectations will adjust accordingly. The forecast also predicts that software engineering outside AI development will slow sharply rather than continue producing broadly adopted new technologies. It argues that developers will have little incentive to create UI libraries when state-of-the-art models are strongest in React, or to create programming languages when those models already know Python, Go, JavaScript, and other established languages best. New libraries and languages may become easier to create, but the prediction is that they will struggle to gain adoption. Large corporations may be exceptions because they can train or fine-tune models on internal technologies, although those technologies could still face difficulty building external communities and talent pools when outside developers lack access to the companies’ models or do not want to use them.
Key Takeaways
- A forecast predicts that AI will handle most software development because lower costs will outweigh reductions in software quality for many businesses.
- The prediction expects consumer and business quality expectations to shift toward commercially sufficient software rather than the highest possible reliability.
- Software engineering is predicted to concentrate on AI development, while new UI libraries and programming languages outside established model knowledge may struggle for adoption.
- Large companies may support proprietary technologies by training or fine-tuning internal models, but could have trouble developing outside user communities and talent pools.
What the discussion said
The discussion focused less on the article’s broad futurism than on whether LLM-assisted programming actually changes what software development is. Some readers see models replacing boilerplate-heavy work, making narrowly tailored code cheaper than adapting a generic framework, and turning established languages and libraries into ever more valuable defaults because models know them best. They also argue that businesses will often choose materially cheaper, imperfect AI-produced software over marginally more reliable human-built systems. Others pushed back that libraries are not merely cognitive shortcuts: they standardize data handling, encode reusable solutions, and let models hand work to cheaper deterministic algorithms. Several commenters likewise rejected the idea that model familiarity freezes innovation. People still create languages and UI frameworks because hard technical problems are intrinsically compelling, though skeptics concede that industrial adoption may concentrate around ecosystems with abundant training data. Current coding agents were described as useful but nowhere near a tenfold productivity leap; reviews, debugging, and ambiguous stakeholder requirements still dominate real delivery. The thread also worried that dependence on hosted LLMs could erode engineers’ core competence and shift control of software production toward model providers. On jobs, participants split between a sharp contraction in routine programming and the historical observation that productivity gains have repeatedly expanded, rather than eliminated, demand for developers.
Where opinion split
The sharp dispute is whether AI coding agents will hollow out routine software engineering or remain another leverage tool that changes the work without eliminating it. Displacement advocates argue that agents already outperform many low-cost contractors on CRUD-style work, making large portions of the workforce economically redundant. Skeptics counter that current gains are modest, requirements discovery and verification remain human bottlenecks, and past automation made software cheaper enough to create more work.
Community Sentiment
Positives
- LLMs can remove repetitive implementation work and make purpose-built code practical where framework conventions once imposed costly compromises.
- Models trained deeply on mainstream stacks could make ordinary application development faster and cheaper for teams that accept imperfect output.
- AI-generated bespoke components may let developers optimize tightly for a single use case instead of bending products around widely adopted libraries.
- Established libraries still help AI systems manage complexity, reuse known solutions, and route suitable tasks to cheaper deterministic methods.
- Better coding assistance could reduce gratuitous framework churn by reinforcing familiar, well-supported ecosystems rather than constantly reinventing them.
Concerns
- Treating libraries as obsolete misses their role in standardized protocols and reusable algorithms, leaving AI-generated systems prone to needless reinvention.
- Model fluency with React, Python, and other dominant stacks may make unfamiliar languages economically awkward, starving better ideas of adoption despite technical merit.
- Current coding agents create a debugging and correction loop rather than an order-of-magnitude speedup, so sweeping replacement claims outrun present capability.
- Routine programmers may lose both jobs and skill-building opportunities if firms substitute agents before engineers develop durable architectural judgment.
- Reliance on hosted LLMs risks handing the means of software production to providers that can charge rent for capabilities teams once owned.
