Skip to main content

Working hypothesis, after a week of peering down the AGI rabbit hole.

Working hypothesis, after a week of peering down the AGI rabbit hole.



I started this blog to chart and comment on the rapid changes that are taking place in the development of narrow AI to see what sort of path we are on, and if there is any possibility that we’d see an emergent AGI, over a course of 6 months.

Any findings are part of an ongoing hypothesis.

Speculations

  • The opportunity for an AGI lies in the distributed links that agent AIs make to perform specific tasks.
  •  No AGI is possible without memory being committed to AI agent results. 
  •  Hardware developments, and the mass deployment of, for example Nvidia DGX H100’s ,will be required for agencies to see what the scale of narrow AIs working in cooperation can bring to more general problems.
  • AI self learning, with the assumption of AI improving upon itself, is unproven. There are many assumptions being made in the AI space.
  • Conflating AI intelligence with animal/human intelligence with sentience remains a stretch at best, hyperbole and misleading at worst.
  • Calling AGI’s ‘God’ even before superintelligence is viable, is not a useful response.
  • Conflating AGIs with popular film fictional representations can be highly misleading.
  • Solving the Alignment issue is unlikely. 
  • Making AGI systems 'fit for purpose', depends upon the purpose. Expectations about purpose will have to be compromised and tailored to existing circumstance.

Legislation

  • Alignment issues will remain problematic, during the course of narrow AI development.
  • Hiding one’s head in the sand about AIs is not a useful response.
  • Legislation as an attempt to control development remains highly problematic; the documents I’ve reviewed from the USA, the EU and some of it’s member states have been inadequate, and have barely been able to understand the problems that AI threw up last year, never mind last month. AI development will now always be too fast for reactionary legislation. 
  • The winners of the productivity changes that AI brings forth will remain with the system elites that created them, so in China for example in their hybrid capitalist system, in the West an increasingly small number of the tech elite. Benefits will be highly constrained for the mass of people, and may well be detrimental to large minorities. This is not speculation.

Tools

  • Current AI tools affect ‘white collar’ work in post industrialised nations the most.
  • Current AI tools remain riddled with biases.
  • Current AI tools have data gatekeepers enacted to present ‘acceptable’ results to consumers of its services.
  • Current AI tools already increase productivity in many white collar fields, to an extent, particularly in more entry level areas, I expect this to grow significantly in the forthcoming days, weeks, months.

Critical Responses

  • We have a catch 22 situation, critical responses without an understanding of the SWOT issues narrow AI and AGI represent are unhelpful, yet keeping up with the developments may not be possible for any human.
  • As ever the range of responses is varied, from tech billionaires calling for a delay in development, under the guise of being helpful, whilst the reality is they are playing catch up, to those who are idolatrous of tech, and to institutions confined by their practicing mission.
Labels:

Comments

Post a Comment

Popular posts from this blog

The AI Dilemma and "Gollem-Class" AIs

From the Center for Humane Technology Tristan Harris and Aza Raskin discuss how existing A.I. capabilities already pose catastrophic risks to a functional society, how A.I. companies are caught in a race to deploy as quickly as possible without adequate safety measures, and what it would mean to upgrade our institutions to a post-A.I. world. This presentation is from a private gathering in San Francisco on March 9th with leading technologists and decision-makers with the ability to influence the future of large-language model A.I.s. This presentation was given before the launch of GPT-4. One of the more astute critics of the tech industry, Tristan Harris, who has recently given stark evidence to Congress. It is worth watching both of these videos, as the Congress address gives a context of PR industry and it's regular abuses. "If we understand the mechanisms and motives of the group mind, it is now possible to control and regiment the masses according to our will without their
Post a Comment
Read more

Beware the Orca, the challenge to ChatGPT and Palm2 is here

  So Google's 'we have no moat' paper was correct. If you train an LLM wisely then it's cost effective and cheap to produce a small LLM that is able to compete or even beat established, costly LLMs, as Microsoft has just found. It's another excellent video from AI Explained, who goes through some of the training procedures, which I won't get into here. Orca, is a model that learns from large foundation models (LFMs) like GPT-4 and ChatGPT by imitating their reasoning process. Orca uses rich signals such as explanations and complex instructions to improve its performance on various tasks. Orca outperforms other instruction-tuned models and achieves similar results to ChatGPT on zero-shot reasoning benchmarks and professional and academic exams. The paper suggests that learning from explanations is a promising way to enhance model skills. Smaller models are often overestimated in their abilities compared to LFMs, and need more rigorous evaluation methods. Explana
Post a Comment
Read more

What is happening inside of the black box?

  Neel Nanda is involved in Mechanistic Interpretability research at DeepMind, formerly of AnthropicAI, what's fascinating about the research conducted by Nanda is he gets to peer into the Black Box to figure out how different types of AI models work. Anyone concerned with AI should understand how important this is. In this video Nanda discusses some of his findings, including 'induction heads', which turn out to have some vital properties.  Induction heads are a type of attention head that allows a language model to learn long-range dependencies in text. They do this by using a simple algorithm to complete token sequences like [A][B] ... [A] -> [B]. For example, if a model is given the sequence "The cat sat on the mat," it can use induction heads to predict that the word "mat" will be followed by the word "the". Induction heads were first discovered in 2022 by a team of researchers at OpenAI. They found that induction heads were present in
Post a Comment
Read more