Considerations:
- Data privacy and security: AI models rely on data, this is oftern proprietary and can contain sensitive information so it is important to implement robust data protection measures.
- Bias and fairness:
- AMAzon model discriminated against female applicants.
- Need to use diverse data sets, implement fairness aware algorithm, audit AI systems for biased outcomes regularly.
- Transparenct and accountability of models.
- Autonomous systems and human oversight
- especially in critical areas like military, health, and transportation
- include human-in-the-loop mechanisms.
- Equitable access: democratize it to bridge the digital divide.
- Environmental impact:
- LLMs are massive and require a lot of computation and electrical and specialized hardware. they have a big footprint.
- Continuous monitoring and improvement.
- Deepfakes:
- videos, images, or audio recordings manupulated using generative AI modls to seem real.
- Fake news/propaganda.
- dpreading disinformation
- harassing or blackmailing people!!!!! damn
- commiting fraud.
- security breaches.
- They can have atrocious consequences if used maliciously: you could fake being the president over the phone and start a nuclear war.
- videos, images, or audio recordings manupulated using generative AI modls to seem real.