AI for Everyone Week 4

The Attitude to AI

Goldilocks rule for AI

Not too optimistic nor too pessimistic

In between view- AI is a powerful tool but has limitations, we can mitigate its potential harms and use it to create tremendous value

Limitations of AI

Performance limitations

Explainability is hard 解释AI的判断结果是一个很有潜力的研究领域

Biased AI through biased data

adversarial attacks on AI

Stereotype

AI的工作原理是找到相对应的向量，Man这个词出现在1，1的位置，给的关联词computerprogrammer的位置在3，2 。 对应词woman出现在（2，3）的位置，AI就会去找Man➡️computerprogrammer 一样的向量，对应在woman为起点的位置上，这样就找到了Homemaker，所以基于现有网络的数据，AI很容易给出充满偏见的指示和意见

Why bias matters

Hiring tool that discriminated against women

Facial recognition working better for light-skinned than dark-skinned individuals

Bank loan approvals

Toxic effect of reinforcing unhealthy stereotype

Combating bias

Technical solutions：

zero out the bias in words 改变词的值到零

use less biased and/or more inclusive data

Transparency and/or auditing processes

Diverse workforce

Adversarial attacks on AI

Minor perturbation - change of the pixels can Fool the AI

DeepFakes-Synthesize video of people doing things they never did

Undermining of democracy and privacy-Oppressive surveillance

Generating fake comment

Spam vs anti-spam and fraud vs anti-fraud

AI其实是很容易被糊弄的，改几个像素，增加个贴纸，就可以让他识别不出停止标志或者识别错误

AI and developing economies

原本的职业发展是像爬梯子一样一阶一阶向上的，但AI的出现会让很多基础级别的岗位得到替代，但同时也会给人们leapfrog，像弹簧床一样直接弹跃到中高级，省去了很多基础的爬坡过程，像是人们不用会编程就可以操作手机和电脑，学校不需要建造教室就可以直接online learning

AI and society

建议使用AI来辅助自己当前具备的专业能力，当然也可以从零开始转做AI

AI会取代很多工作，同时也会创造很多新的工作

要保持终身学习

大学学四年 工作四十年的日子一去不复返了