Take AISafety.info’s 3 minute survey to help inform our strategy and priorities

Take the survey
Objections and responses

Is smarter-than-human AI unrealistic?
Is AI alignment easy?
Flawed proposals for setting AI goals
Flawed proposals for controlling AI
Dealing with misaligned AGI after deployment
Other issues from AI
Morality
Objections to AI safety research
Miscellaneous arguments

But won't we just design AI to be helpful?

One might think that aligning an AGI

would be easy. Companies are doing a decent job at preventing LLMs such as GPT-4 from generating harmful text; why would this not continue as AI gets more and more powerful? Isn't there a simple fix that allows us to get robust alignment?

Current AI models might be sufficiently aligned to be useful, but despite the best efforts of the labs that produce them, they still suffer from issues such as bias and susceptibility to adversarial prompts. Furthermore, resolving these issues may not be enough: there are reasons to expect that, as AI systems become more capable, systems that seem to display aligned behavior in some environments could turn out to be dangerously misaligned in others. Dive into the related articles to see what makes AI alignment

a hard problem and why current alignment techniques might not scale to more powerful AI.

Keep Reading

Continue with the next entry in "Objections and responses"
Aren't there easy solutions to AI alignment?
Next
Or jump to a related question


AISafety.info

We’re a global team of specialists and volunteers from various backgrounds who want to ensure that the effects of future AI are beneficial rather than catastrophic.