AI SECURITY
“ AIs do not have rights, feelings or consciousness. And we must not train them to act as though they do”
Mustafa Suleyman CEO Microsoft AI
“ It’ s easy to see how an entity trained in this way would act like it is entitled to certain freedoms, protections and rights. And it’ s hard to imagine how we could control such an entity,” Mustafa notes.
This difference in approach perhaps requires a debate, while what does not is the fact that research needs to slow down, alignment needs to be better studied and that industry regulation and guardrails are mandatory. In the current scenario, where Jacob told the Wall Street Journal that“ we are on track for a lot of the most aggressive scenarios where by the end of next year, things could be out of control already,” caution is in fact the right decision.
In light of this, OpenAI recently released a“ framework for tracking, investigating and disclosing instances of model misalignment”. Included along with this, were reports of six instances of misalignment observed at OpenAI. Some of the new and prominent cases are examined in detail here.
cybermagazine. com 101