AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
The term AGI, or artificial general intelligence, was coined in the early 2000s by a group of researchers working on creating thinking machines that could rival human-level intelligence. The term AGI was chosen because it was immediately comprehensible to people with a technical background and it emphasized the goal of achieving general human-level intelligence.
The AGI community has been around for several decades, with researchers and enthusiasts dreaming of creating super-intelligent systems. While the term AGI may not have been widely used until recently, there has always been a group of individuals pushing for the development of AGI. These pioneers laid the groundwork for what we see today, and they fostered a sense of community among those dedicated to the pursuit of AGI.
The AGI community has grown in recent years, with conferences and workshops dedicated to the discussion of AGI and its implications. While still a relatively small community compared to other branches of AI, the AGI community has attracted the attention of national leaders and corporate CEOs. However, it remains distinct from mainstream AI research, with a focus on creating truly general and human-like intelligence.
OpenCog is an open-source software platform that utilizes a weighted labeled hypergraph, known as the AtomSpace. This platform allows for the representation of various types of knowledge, such as logic-like nodes and links, neural-like nodes and links, and procedure-like nodes and links. The AtomSpace serves as a knowledge store, and the platform includes utilities like pattern matching and scheduling agents to operate on the hypergraph.
OpenCog is exploring ways to improve interoperability with computational graphs of neural networks. Efforts are being made to develop faster type checking among dependent types and probabilistic dependent types. The goal is to efficiently integrate deep neural net frameworks like Torch with OpenCog's hypergraph, allowing for more seamless cooperation and interaction between different types of learning algorithms.
OpenCog is also engaged in ongoing research on unsupervised grammar induction. This involves teaching an AI system to learn the grammar of a language by examining a large corpus of text. OpenCog combines symbolic grammar learning algorithms with the use of transformer neural networks as sentence probability oracles, allowing for more effective grammar induction and learning.
OpenCog is currently undergoing a major rebuild to create the True AGI system. The emphasis is on enhancing scalability, speed, and efficiency, with a focus on distributed processing and improved execution of dependent types. The core AGI paradigm remains intact, but these improvements aim to address performance requirements and optimize the overall functionality of the system.
Sophia, created by David Hanson, is a robot designed to evoke emotions and empathy from people through its humanoid appearance. The goal is to create a platform for beneficial applications and to infuse the AI with love and compassion. While some in the AI community may criticize Sophia for being mistaken as AGI, the intention is to utilize the human-like connection to provide a positive user experience and teach the AI about human values. Although Sophia's capabilities are often attributed more intelligence and consciousness than it possesses, being upfront about how it works has been a priority, even though people still tend to ascribe more to her. This ethical approach aims to enhance human-robot interactions and fulfill helpful roles in areas such as nursing, education, and elder care.
Viruses can be seen as complex, self-organizing adaptive systems. While the concept of intelligence may not seamlessly apply, similarities exist in their adaptive capabilities. Human immune systems, also complex and adaptive, engage in an ongoing interaction with viruses. In the fight against COVID-19, machine learning and machine reasoning techniques are being employed to assist doctors in personalized medicine. The goal is to determine the most effective combination of antivirals based on genomics and clinical indicators. The challenges lie in data integration, transfer learning, and prediction accuracy. This approach showcases the potential of AI and data analytics to combat and understand the virus while paving the way for personalized treatments.
Access to real-time, open-source COVID-19 clinical trial data would greatly benefit analysis, prediction, and personalized medicine. However, current data silos and restrictions hinder progress in this area. The ideal scenario would involve making data accessible for analysis and interpretation with proper privacy protection. Part of the challenges lie in political and geopolitical factors, as cooperation between countries like the US and China may be affected. However, initiatives are underway to leverage AI, such as machine learning and reasoning, to analyze COVID-19 data sets and support medical decision-making. The utilization of decentralized blockchain platforms like SingularityNet further allows for secure sharing and analysis of data, providing opportunities for collaborative research against future pandemics.
The speaker emphasises the importance of prolonging life and defeating death through advancements in longevity research and technology. They believe that death is not a positive aspect of life and that people should not die unless they choose to. The goal is to abolish involuntary death and create a future where individuals have the opportunity to live indefinitely with the ability to choose when and if they want to end their own lives. The speaker mentions various potential approaches to prolonging life, such as stem cell injections and targeted molecular interventions, and highlights the need for extensive funding and research into these areas.
The speaker discusses the significance of artificial general intelligence (AGI) and its potential impact on humanity. They argue that upgrading human intelligence to superintelligence would allow for the exploration of different variations of joy, experiences, and values, going beyond the limitations of the human form. A superintelligent AGI that embraces compassion and human values is seen as an essential element in creating a beneficial technological singularity. To achieve this, the speaker advocates for integrating AI and biological knowledge, decentralizing research, fostering open collaboration, and addressing ethical and regulatory challenges.
Ben Goertzel is one of the most interesting minds in the artificial intelligence community. He is the founder of SingularityNET, designer of OpenCog AI framework, formerly a director of the Machine Intelligence Research Institute, Chief Scientist of Hanson Robotics, the company that created the Sophia Robot. He has been a central figure in the AGI community for many years, including in the Conference on Artificial General Intelligence.
Support this podcast by supporting these sponsors:
– Jordan Harbinger Show: https://jordanharbinger.com/lex/
– MasterClass: https://masterclass.com/lex
This conversation is part of the Artificial Intelligence podcast. If you would like to get more information about this podcast go to https://lexfridman.com/ai or connect with @lexfridman on Twitter, LinkedIn, Facebook, Medium, or YouTube where you can watch the video versions of these conversations. If you enjoy the podcast, please rate it 5 stars on Apple Podcasts, follow on Spotify, or support it on Patreon.
Here’s the outline of the episode. On some podcast players you should be able to click the timestamp to jump to that time.
OUTLINE:
00:00 – Introduction
03:20 – Books that inspired you
06:38 – Are there intelligent beings all around us?
13:13 – Dostoevsky
15:56 – Russian roots
20:19 – When did you fall in love with AI?
31:30 – Are humans good or evil?
42:04 – Colonizing mars
46:53 – Origin of the term AGI
55:56 – AGI community
1:12:36 – How to build AGI?
1:36:47 – OpenCog
2:25:32 – SingularityNET
2:49:33 – Sophia
3:16:02 – Coronavirus
3:24:14 – Decentralized mechanisms of power
3:40:16 – Life and death
3:42:44 – Would you live forever?
3:50:26 – Meaning of life
3:58:03 – Hat
3:58:46 – Question for AGI
Listen to all your favourite podcasts with AI-powered features
Listen to the best highlights from the podcasts you love and dive into the full episode
Hear something you like? Tap your headphones to save it with AI-generated key takeaways
Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more
Listen to all your favourite podcasts with AI-powered features
Listen to the best highlights from the podcasts you love and dive into the full episode