Do we know what is at stake?

Do we know what is at stake?

You will have seen from my posts, that I side with Ilya Sutskever (head of research) in this OpenAI struggle. Like him, I think he made the wrong decision to vote with the OpenAI Board to remove Sam Altman. I've read funny posts that show how if the OpenAI Board had asked ChatGPT what they should do, they would have received better advice on what to do! I side with Ilya because he realises what is at stake here. He realises the path that Sam and Greg were taking OpenAI down.


The reason OpenAI exists is to make sure that AGI (Artificial General Intelligence) when it arrives, makes life better for all humanity. An AGI unaligned to #AIforGood will be catastrophic. My view is that the biggest risk of AI / AGI is not limited to "bad people", but rather more likely to be the unintended consequences of the use of very powerful AI/AGI.

It appears that in the last few weeks, based on Sam Altman's public comments, that OpenAI researchers led by Ilya have developed a powerful new step towards AGI. Sam described that we will see GPT-4 Turbo as 'quaint' by this time next year. Powerful AI mustn't be released without the necessary work baking in Alignment.

This is important. AGI is likely to develop when AI has a capability similar to or greater than AI developers. At that stage, it will be able to use simple compute power to improve itself and there starts the potential for AI development without guardrails. It could be months, weeks or days before the AI develops into something we might define as AGI. If this has no alignment then the risks from unintended consequences become unknown.

From what I am reading at the moment, people are seeing the OpenAI troubles as similar to other corporate breakdowns. In this case, I believe it is not and it is much more than that.

Does Ilya have a choice he can make that safeguards the development of AGI? I think the best he can hope for is keeping OpenAI's mission viable. Achieving that seems more complex by the day.

*I didn't use generative AI to write this, every word is my own.

Alex Ferguson

AI | Solution Architect | Developer | Advisory Board Member

1 年

Ilya, after a failed Coup attempt, suffers from an immediate dose of regret… not surprising.

回复
Christopher Izquierdo ??

Premium Ghostwriter | I ghostwrite educational email courses for Functional Medicine and Corporate Wellness Founders. 15+ years Nurse Leader and obsessed with AI strategy

1 年

Thanks Matt, very balanced point of view. I definitely agree. Ilya is a scientist first of all. And would have made the difficult decision to oust Altman with safety in mind. But at the end of the day innovation and AI advancement won the game. Which we hope aligns with safe exploration and implementation of AGI.

Matt Ensor

CMInstD, MEngNZ, Chair - AI Forum GenAI WG

1 年

Here we go, introducing Q*

  • 该图片无替代文字
回复
Matt Ensor

CMInstD, MEngNZ, Chair - AI Forum GenAI WG

1 年

The announcement of the return of Sam Altman to CEO role of OpenAI secures their mission again, but potentially with doubts about whether safety and alignment is being held up as essential prior to progress. Ilya’s role needs to stay on the Board in my view as it is likely that Microsoft will regain a Board representative.

回复

要查看或添加评论,请登录

Matt Ensor的更多文章

  • Govt Ministers need this now

    Govt Ministers need this now

    This morning I installed a fully functional Large Language Model (LLM), Generative AI, on my iPhone. It doesn't need a…

    2 条评论
  • Living in Lala Land

    Living in Lala Land

    As a board member, shareholder, and participant in various professional circles, I’m still startled that in many…

    3 条评论
  • Beware AI oversharing secrets

    Beware AI oversharing secrets

    I'm fortunate to have both FranklyAI for Microsoft Teams and CoPilot for Microsoft Office on my work computer, and they…

    4 条评论
  • Will the legal world be able to compete?

    Will the legal world be able to compete?

    I added a legal contract due diligence capability into FranklyAI to see how our in-house legal team would react. I…

    1 条评论
  • It's time to replace the TMO

    It's time to replace the TMO

    I want to propose something radical, let's replace the TMO with GPT-4. We won't need to wait more than a few seconds…

    2 条评论
  • Jumbo Jets don't fly do they?

    Jumbo Jets don't fly do they?

    I vividly remember being on the observation deck at Auckland Airport with my father for the first time watching a…

    2 条评论
  • Let's break your rules

    Let's break your rules

    I've been curious about AI Ethics for a long time, especially so when the ChatGPT emerged in November 2022. Many of you…

  • This is(n't) safety.

    This is(n't) safety.

    So, there I was, attending the TRAFINZ 2023 conference when a slide by Professor Narelle Haworth on micromobility…

    5 条评论
  • Brace for impact.

    Brace for impact.

    If generative AI is so ground-breaking why is it being ignored by the majority of humans? The socio-psychological…

    10 条评论
  • This is a warning.

    This is a warning.

    In years gone by I have enjoyed playing RISK, a board game where you attempt to take over the world. The best strategy…

    14 条评论

社区洞察

其他会员也浏览了