跳到内容

Risks & Pain Points

The AI industry has never been shy about discussing the technology's potential for disaster, sadly stopping just short of actually, you know, slowing down for a minute.

Said risks include job displacement as a top concern, but also AI's potential for military use as seen in recent conflicts, and of course the worrying environmental impact of AI in a world already threatened by climate change.

It's also interesting to note that women and non-men respondents overall were much more likely to worry about AI's environmental impact (58% vs 37%).

Many commenters complained about being limited to selecting only three items. As one person put it: “It's kind of shocking how big this list of negative social impacts are when you put them all in one place.”

Which of these general AI risks and issues are you most concerned about?
多选
Limited
3
0%
20%
40%
60%
80%
100%
01
Job displacement
02
Military use of AI
03
Environmental impact
04
AI slop takeover
05
Negative cognitive impacts
06
Security issues
07
AI overuse
08
Rising AI costs
09
Copyright and ethics issues
10
Open-source disruption
0%
20%
40%
60%
80%
100%
-17%
Percentage decrease in respondents who cited prompt comprehension as a pain point.

When it comes to issues with actually using AI, Hallucinations lead the pack.

What makes this issue uniquely frustrating is not just having to deal with incorrect results, but incorrect results confidently presented as correct, with the model having no awareness of its own potential for error.

在使用 AI 工具时,您最大的痛点是什么?
多选
Limited
3
0%
20%
40%
60%
80%
100%
01
幻觉和不准确
+8
02
代码质量
-1
03
缺乏上下文
-1
04
财务成本
+3
05
隐私担忧
-1
06
提示词理解
-3
07
版权担忧
-2
08
集成困难
+1
09
响应时间慢
10
道德伦理担忧
11
缺乏对提示词的理解
12
生成的代码质量差
13
其他答案
0%
20%
40%
60%
80%
100%

缺失的功能

Confirming the previous question's results, Truthfulness is the leading missing feature. In an age where we're growing increasingly skeptical of our fellow humans' ability to distinguish fact from fiction, it'd be nice if our AI at least wouldn't gaslight us as well.
您觉得 AI 工具目前缺少哪些功能?
多选
Limited
3
0%
20%
40%
60%
80%
100%
01
真实性
02
Long-term memory
03
最新的知识
04
Accountability
05
Awareness of runtime behavior
06
High-level planning
07
更大的上下文窗口
08
更好的 UI/UX
09
支持更大的代码库
10
更好的工具集成
0%
20%
40%
60%
80%
100%
Overall happiness with AI hasn't changed much since last year, another sign that developers might still be in a “wait and see” holding pattern.
您对 Web 开发 AI 工具的现状满意吗?
1
2
3
4
5
2025
2026
1
2
3
4
5
平均 {axis}
MCP: Learn the Universal Language for AI Integration
Brian Holt
Neon

MCP: Learn the Universal Language for AI Integration

Dive into the Model Context Protocol and code an MCP server from scratch! Leverage external APIs, interact with custom applications, and learn security best-practices to ensure safe AI agent deployment.
Hard Parts of Neural Networks
Will Sentance
Codesmith

Hard Parts of Neural Networks

Through hands-on examples such as fraud detection and image recognition, you'll build robust mental models for foundational AI concepts like data preprocessing, probability, training, and prediction.
感谢合作伙伴对我们的支持! 了解更多。