The Incentive to Deceive

Rob Miles is one of the better explainers of AI on YouTube, he's detailed, he rarely holds back on calling out elephants, and, importantly for broadcast media, he's personable. He's also has a long, in YouTube terms, track record of covering Alignment issues. As a PhD student he's particularly adept at explaining the complexities of Alignment issues. In this video he gives a fine explanation of the reward training in LLM's both implying and stating the issues that ensue from such training approaches, including the policies to please humans, and the utility of such models to deceive.

Two parts near the end of the video caught my attention:

'This is potentially fairly dangerous, there are certain type of goals that are instrumentally valuable for a wide range of different terminal goals, in the sense that, you can't get what you want if you're turned off, you can't get what you want if you're modified, you probably want to gain power and influence."

'Reinforcement Learning From Human Feedback, is a powerful Alignment technique, in a way, but it does not solve the problem...extremely powerful systems trained in this way, I don't think they'd be safe.'

Comments

The Whispers in the Machine: Why Prompt Injection Remains a Persistent Threat to LLMs

Large Language Models (LLMs) are rapidly transforming how we interact with technology, offering incredible potential for tasks ranging from content creation to complex analysis. However, as these powerful tools become more integrated into our lives, so too do the novel security challenges they present. Among these, prompt injection attacks stand out as a particularly persistent and evolving threat. These attacks, as one recent paper (Safety at Scale: A Comprehensive Survey of Large Model Safety https://arxiv.org/abs/2502.05206) highlights, involve subtly manipulating LLMs to deviate from their intended purpose, and the methods are becoming increasingly sophisticated. At its core, a prompt injection attack involves embedding a malicious instruction within an otherwise normal request, tricking the LLM into producing unintended – and potentially harmful – outputs. Think of it as slipping a secret, contradictory instruction into a seemingly harmless conversation. What makes prompt inj...

Podcast Soon Notice

I've been invited to make a podcast around the themes and ideas presented in this blog. More details will be announced soon. This is also your opportunity to be involved in the debate. If you have a response to any of the blog posts posted here, or consider an important issue in the debate around AGI is not being discussed, then please get in touch via the comments. I look forward to hearing from you.

The tech utopia of endless leisure time is here: goodbye jobs

'AI eliminated nearly 4,000 jobs in May' so it's reported by hallenger, Gray & Christmas, Inc. Following on from reports by IBM et al that thousands of job cuts will occur due to AI replacement, there is no need to wait for the utopia of AI allowing humans more leisure time, as that's already here, in the form of redundancies, if we are to accept the reports findings. 'With the exception of Education, Government, Industrial Manufacturing, and Utilities, every industry has seen an increase in layoffs this year.' What's particularly notable is that it's the Tech sector that's the most affected from job cuts in the US economy: 'The Technology sector announced the most cuts in May with 22,887, for a total of 136,831 this year, up 2,939% from the 4,503 cuts announced in the same period last year. The Tech sector has now announced the most cuts for the sector since 2001, when 168,395 cuts were announced for the entire year. ' Another reason ...

Charting the emergence of AGI?

Search This Blog