AGI: An Artificial Intelligence That Could Exceed Human Intelligence
Last time I wrote about AI could replace some types of developers. While technologies such as ChatGPT 3.0 are highly capable to code, they still need very specific instructions. We have to add that ChatGPT 3.0 was modeled with data that was processed until 2021, and there are several new technologies that AIs such as ChatGPT 3.0 could not successfully process or code, even providing incorrect information. Human interaction still being dependable when using these technologies. And currently, it just could increase the performance, instead of totally substituting. (Unless you are the very specific developer that I mentioned in the previous article).
However, did you know that there is something smarter than the narrow AI we have today and that AI could have the intelligence compared to humans?
Well, I introduce you to AGI.
AGI: Artificial General Intelligence
AGI, short for Artificial General Intelligence, is a term that does not have a precise definition due to ongoing debates within the field. However, at its core, AGI aims to replicate the remarkable problem-solving abilities and cognitive functions exhibited by humans. This involves enabling AI systems to autonomously find solutions, adapt to new situations, and possess a broad understanding across various domains.
In simpler words, AGI seeks to emulate human cognition.
What is cognition?
“Simply put, cognition is thinking, and it encompasses the processes associated with perception, knowledge, problem-solving, judgment, language, and memory.” ⁴
Practically, the way we process a task, and find a solution for that task, using our reasoning, memories, and creative problem-solving. In essence, AGI seeks to capture the essence of human cognition.
But ChatGPT 3.0 does this already, right?
Well, the answer is no. ChatGPT, at least the 3.0 version, is not there yet. For example, ChatGPT 3.0 does not have “memory”. According to the Community Forums of OpenAI,³ each request sends the entire conversation history for each request, so it does not exactly “remember”. It also does not have reasoning capabilities, consciousness, or even a true understanding of the information that is being processed.
It just finds patterns of what it learned from the training data and provides a human-like response.
And there is when AGI comes…
AGI on the other hand, will have the ability to understand what is the task, and reasoning, learn from it, and apply a solution. And yes, it is comparable with human intelligence.
Many experts say it could have more intelligent “minds” than humans do. One of them is Max Tegmark, a professor doing AI research for MIT, in an article written for “Time” ⁸, which considers AGI to be a threat to humans.
Another expert considered the AI ‘godfather’, Geoffrey Hinton, told BBC the AI Dangers ⁵. Similarly, Max Tegmark, shares the view that AI could be more intelligent than us, and as he describes, it could be “quite scary” ⁵. He provided examples of the bad guys using AI to do bad things. For this case, Geoffrey Hinton was referring to AI bots specifically.
The problem here? Basically, humans would have to compete with superior intelligence.
However, we have OpenAI CEO, Sam Altman, which seems to look at AGI with a more positive outlook: “make people smarter and healthier”. ²
How long we should wait to see AGI?
Experts such as Demis Hassabis, CEO of Deepmind, for Wall Street Journal, share his views and told that AGI could be possible in a few years, "at least in some form. " ⁶
According to AI Impacts, a 2022 Expert Survey on Progress in AI ¹, the majority of surveyed estimate that high-level machine intelligence will be reached around 2059. (High-level machine intelligence is often interchangeable with AGI).
On April 13, 2023, Microsoft published research about the “Sparks of Artificial General Intelligence” ⁷, demonstrating that AGI could be here sooner than expected and that ChatGPT 4.0 could show signs of AGI.
In summary…
In this article, we have examined the fundamental concepts surrounding Artificial General Intelligence (AGI) and delved into its current state. AGI embodies the quest for autonomous systems that can mimic human intelligence. While a precise timeline for its full realization remains elusive, the remarkable advancements in AI research in recent years provide a promising trajectory toward its eventual emergence.
The implications of AGI on society are both captivating and thought-provoking. As we anticipate its arrival, we face a future filled with possibilities and uncertainties. The impact on various aspects of our lives, for better or worse, will undoubtedly be profound and intriguing.
References
1. AI Impacts. (2022, August 3). 2022 Expert Survey on Progress in AI. Retrieved from https://aiimpacts.org/2022-expert-survey-on-progress-in-ai/#High-level_machine_intelligence_timelines
2. Altman, S. [@sama]. (2023, May 7). “here is an alternative path for society: ignore the culture war. ignore the attention war. make safe agi. make fusion. make people smarter and healthier. make 20 other things of that magnitude. start radical growth, inclusivity, and optimism. expand throughout the universe.” [Tweet]. Twitter.
3. Armstrong, P. (2023, April 11). Response to “It will be much better, when chat GPT can remember about all previous chats.” Retrieved from https://community.openai.com/t/it-will-be-much-better-when-chat-gpt-can-remember-about-all-previous-chats/152017
4. Dumper, K., Jenkins, W., Lacombe, A., Lovett, M., & Perimutter, M. (n.d.). Simply put, cognition is thinking. Retrieved from https://opentext.wsu.edu/psych105/chapter/7-2-what-is-cognition/
5. Kleinman, Z., & Vallance, C. (2023, May 2). AI ‘godfather’ Geoffrey Hinton warns of dangers as he quits Google. BBC News. Retrieved from https://www.bbc.com/news/world-us-canada-65452940
6. Kruppa, M. (2023, May 2). Google DeepMind CEO Says Some Form of AGI Possible in a Few Years. Retrieved from https://www.wsj.com/articles/google-deepmind-ceo-says-some-form-of-agi-possible-in-a-few-years-2705f452
7. Sebastien Bubeck, Varun Chandrasekaran, Ronen Eldan, Johannes Gehrke, Eric Horvitz, Ece Kamar, Peter Lee, Yin Tat Lee, Yuanzhi Li, Scott Lundberg, Harsha Nori, Hamid Palangi, Marco Tulio Ribeiro, & Yi Zhang. (2023, April 13). Sparks of Artificial General Intelligence: Early experiments with GPT-4. Retrieved from https://arxiv.org/pdf/2303.12712.pdf
8. Tegmark, M. (2023, April 25). The ‘Don’t Look Up’ Thinking That Could Doom Us With AI. Retrieved from https://time.com/6273743/thinking-that-could-doom-us-with-ai/
Do you identify as Latinx and are working in artificial intelligence or know someone who is Latinx and is working in artificial intelligence?
- Get listed on our directory and become a member of our member’s forum: https://forum.latinxinai.org/
- Become a writer for the LatinX in AI Publication by emailing us at publication@latinxinai.org
- Learn more on our website: http://www.latinxinai.org/
Don’t forget to hit the 👏 below to help support our community — it means a lot!
Thank you :)
