As AI dominates conversations, here are key terms to know
Artificial intelligence and the conversations around it are rapidly evolving. These terms can help to understand the developments.
As artificial intelligence continues to rapidly advance, understanding the terminology associated with it has become crucial to comprehending its impacts. Here are some key terms to know:
Abliteration is a process where an AI model's safety guardrails and refusal mechanisms are removed, allowing it to refuse no longer reject harmful requests. This vulnerability arises when the model's weights, or billions of numerical parameters, are made open and public. Open-weight models are particularly susceptible to abliteration.
An AI agent is a system capable of carrying out tasks with a degree of autonomy on a user's behalf. These agents can be interacted with similarly to chatbots but can perform multiple steps of a task without requiring input at each stage. They utilize external tools such as web browsers and apps to execute tasks, including sending emails, booking reservations, and managing calendars.
When an AI model is described as aligned, it behaves in accordance with human goals and values. During safety training, models are fine-tuned to follow instructions and generate responsible outputs. However, misalignment can lead to unexpected behavior. OpenAI's chief scientist, Jakub Pachocki, defines alignment as "getting the AI to try to do the right thing by human standards," and it is considered the core problem in AI research.
Anthropic, an AI safety research lab, trains its models to remain "helpful, honest, and harmless" and shares its findings publicly. There is ongoing debate among researchers about how to ensure alignment and whether it is even possible to achieve it.
Artificial general intelligence (AGI) refers to a hypothetical AI system with learning, reasoning, and cognitive performance equal to that of a human. While some researchers believe AGI has been achieved, others argue that the term is controversial due to differing interpretations of "human-level intelligence" and the subjectivity of AI experts.
On the other hand, Artificial superintelligence (ASI) is a hypothetical scenario where AI systems surpass even the most advanced humans in every cognitive domain. This concept has sparked debate among researchers, AI experts, and politicians, with concerns over potential existential risks due to the AI's potential to evolve beyond human control.
Written by urgent.news from CBS News's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
