www.lesswrong.com/posts/HekpnSkrt89tMm3Dc/i-am-an-ai-safety-researcher
1 correction found
continued to work on both building such an AI and ensuring its successors would remain aligned to the original
MIRI’s own history says it stopped pursuing AGI-building very early on. Around 2003 it pivoted to alignment research, and its later work is described as field-building and alignment research rather than building AI systems.
Full reasoning
MIRI’s official history contradicts the claim that, throughout the 2000s and 2010s, it was both building AI and aligning it.
According to MIRI’s own 2024 strategy update, the organization was founded in 2000 with the goal of accelerating smarter-than-human AI, but "shifted its focus to the alignment problem around 2003." The same page describes its later periods as:
- ~2006–2012: building communities and fields of inquiry around existential risk and related topics
- starting in 2013: "Agent Foundations research" and helping the external alignment field get started
- 2017–2020: a new "engineering-heavy set of alignment research directions"
Likewise, MIRI’s public “About” page says that after 2000–2003, it "pivoted to focus most research efforts on ASI alignment", then spent 2013–2018 on "foundational mathematical problems relevant to AI alignment".
So it is inaccurate to say MIRI "continued to work on both building such an AI and ensuring its successors would remain aligned" across the 2000s and 2010s. MIRI’s own timeline says it stopped pursuing AGI-building as an organizational focus around 2003 and spent the following years on alignment and field-building instead.
2 sources
- MIRI 2024 Mission and Strategy Update - Machine Intelligence Research Institute
When MIRI was first founded ... in 2000, its goal was to try to accelerate to smarter-than-human AI as quickly as possible... Eliezer came to the conclusion that he was completely wrong ... and MIRI shifted its focus to the alignment problem around 2003. ... In ~2006-2012, our primary focus was on trying to establish communities... Starting in 2013, our focus was on Agent Foundations research... In 2017-2020, we shifted our primary focus to a new engineering-heavy set of alignment research directions.
- About the Machine Intelligence Research Institute
2000-2003 ... MIRI pivoted to focus most research efforts on ASI alignment ... 2013-2018 Research focused on foundational mathematical problems relevant to AI alignment.