Jump to content

User:Regulinecoast1/Artificial general intelligence

From Wikipedia, the free encyclopedia
DeepBlue, the artificial system used to defeat Gary Khasperov. T

What is AGI?

As of 2021, scientists are working on something that was only seen in a person's favorite science fiction movies, artificial general intelligence. Commonly known as AGI, Artificial General Intelligence is the idea that a machine mimic a person's thinking and actions. This type of intelligence was popularized within the sci-fi genre of books and movies. Common themes of AGI includes communicating like humans, think like humans with the bonus of being a being who has the potential to do tasks humans cannot achieve thanks to being computed in that liking. The idea of Atifical General Intelligence is still not concrete, but there is promise within this type of AI. AI can now be used in situations where people were doing the work before, such as automated chat-boxes instead of going to the HR department and recomendation systems. [1]


Background of Artificial Intelligence

AlphaGo, a reactive machine, is playing a human Go player Lee Sedol.

Artificial Intelligence has been used or explored by scientists and engineers for quite some time now. Scientists explored the notion of having machines having some like minded intelligence seen in humans. However, AI has been deemed as too radical in unique cases. An example for this case is seen in 1997. A man named John Mccarthy critiqued DeepBlue, a supercomputer that defeated a chess world champion, Garry Khasperov. Mccarthy then stated that the computer anticipated plays that a normal player cannot predict. [2] On the other hand, Khasperov argued that the computer played too similar to other human players of chess.[3] As more time has passed, other feats of artificial intelligence have been done in order to advance technology performance.

One story remarks a research team from Stanford University in California during 2005. The team manufactured a Volkswagen that did not rely on any form of human interaction. The automobile was guided by a satelitte. The car then traversed through a course in the desert in 6 hours an 25 minutes.[4]

Artificial Neuron Network, artificial intelligence used in the medical field today.


What is Artificial Intelligence?

Artificial Intelligence at its core is the mimicking of the human brain to think like humans and copy their actions. The main goal for artificial intelligence in the future is to take action for a specific goal. There are four types of AI that accomplish this goal.

Reactive Machines

Reactive machines is seen as the most basic form of AI. Reactive machines are bilt to react to situations that they are out in. They are not built to remember a task or stores past data. Deep Blue is an example of this technology. The computer could react to what Gary Khasperov was doing based on predictions alone. However, the computer can only do this. It can not remember what Khasperov did and then use his move in another game. Reactive machines are mostly seen within specific niches like video games or Go as seen with Google's AlphaGo.[5]


Limited Memory

Limited Memory is an AI that gain knowledge based on previous experiences. The one major difference between a Reactive Machine and a machine with Limited Memory is that a limited memory machine observes the information that is sent to them. [6] On the other hand, reactive machines cannot call back a prediction that they had just did. This is a process that is important for artifical general intelligence. AGI would need to observe and recall information that is sent to them in order to have the functionality of a human.

Theory of Mind

Theory of mind is the theory that machinery will interact similarly to how humans interact on a day to day basis. While there are some voice assistance machines out in the world, there is no technology as of right now that can converse the way that human can. Speaking is a key separator between humans and robots. In order for robots to have this concept, robots have to converse with the emotions, tone, and variations that all humans can do. Robots must also be able to understand social situations, take the cue and then adapt to to modern life like humans have done. What makes theory of mind so hard to develop is that scientists have not figured out how the human brain can be implemented to a computer. Until neuroscientists have figured out how to, this will most likely be almost impossible to achieve.

Self Awareness

The fourth and most advanced form of artifical intelligence is self-awareness. Self awareness in an AI is when artifical intelligence has reached human levels of consciousness. There is no current machine that has reached this level of AI yet. Self aware AI have the ability to have actions comparable to humans like theory of mind AI. Self-aware AI focus on having self guided thoughts and reflection . They have the ability to understand feelings from themselves and others. Similarly, it is almost impossible to make this intelligence a reality as of this year. [7]


Modern AI and Artificial General Intelligence

In modern times, AI is used in many different areas of work and studies. One profession AI is used heavily is in the medical field. Researchers have developed ANN, Artificial Neuron Networks. ANN mimics the brain's problem solving prowess. ANN has hundreds of artificial nodes that help process data and show results from previous patients or appointments. [8]ANN has solved many things that humans had difficulty solving or not solving at all. ANN has been used by many doctors in many fields of medicine, especially in radiology. A study has been done by Sarah Burdett et al with a ProstAcute Index which scans for prostate cancer and the condition of the prostate. The machine had a 90% accyracy rate. Other applications ANN is used for is abdominal pain, back pain, and other conditions like glacuoma. [Department of Computer Science, University of Hull, UK 1] [9]

Robots is seen as a version of artificial general intelligence. Robots can lift heavy loads and do tasks that humans cannot do. However, it has trouble with tasks that humans find easy, like communication. Robots are being explored as a way to become more efficient at distributing a product, as they can be engineered to pick up objects at a rate that humans cannot achieve.[10] We are seeing more and more exploration of robots and other test runs in order to make manual labor easier for bigger companies who can afford the cost.

Companies like Amazon have over 200,000 robots under their command in Massachusetts. As of now, Amazon is still researching and developing robotic processes in order to automate their shipping services.

The road to having AGI is coming faster than one might expect. Many experts believe that there will be a breakthrough for AGI in 2030. Around 2060 is when technology experts are making AGI that passes a consciousness test. [11] There are currently over 40 organizations that are working on AGI. Their goals with AGI is to have AGI be built to benefit intellectual purposes, animals, ecosystems, advances biological organims and other beings which includes the AGI, and the human race as a whole.[12]


[13]










References

[edit]
  1. ^ Walsh, Kathleen. "Is Artificial General Intelligence (AGI) On The Horizon? Interview With Dr. Ben Goertzel, CEO & Founder, SingularityNET Foundation". Forbes. Cognitive.
  2. ^ Dick, Stephanie. "Artificial Intelligence".
  3. ^ Anderson, Mark. "Twenty years on from Deep Blue vs Kasparov: how a chess match started the big data revolution". TheConversation.
  4. ^ Markoff, John. "Behind Artificial Intelligence, a Squadron of Bright Real People". New York Times.
  5. ^ Hintze, Arend (November 14, 2016). "From Reactive Robots to Sentient Machines: The 4 Types of AI". LiveScience. Retrieved 23 April 2021.
  6. ^ Raynoso, Rebecca (27 March 2019). "4 Main Types of Artificial Intelligence". Learn G2. Retrieved 2 May 2021.
  7. ^ Fonseca, Lauren (4 May 2021). "An Introduction to Artificial Intelligence: The Four Types of AI". The Difference Engine. Retrieved 5 May 2021.
  8. ^ Frankenfield, Jake (August 28, 2020). "Artificial Neuron Network". Investopedia. Retrieved 23 April 2021.
  9. ^ "Prostate Radiotherapy for Metastatic Hormone-sensitive Prostate Cancer: A STOPCAP Systematic Review and Meta-analysis". ScienceDirect.com. Retrieved 11/March/2021. {{cite web}}: Check date values in: |access-date= (help)
  10. ^ "How Artificial Intelligence Works" (PDF). European Parliament. Retrieved 11/March/2021. {{cite web}}: Check date values in: |access-date= (help)
  11. ^ English, Trevor (April 30, 2020). "How Close We Are to Fully Self-Sufficient Artificial Intelligence". InterestingEngineering.
  12. ^ Baum, Seth (16 Nov 2017). "A Survey of Artificial General Intelligence Projects for Ethics, Risk, and Policy". GCRINInstitute. Global Catastrophe Risk Institute. Retrieved 23 April 2021.
  13. ^ Edwards, David. "Amazon now has 200,000 robots working in its warehouses". roboticsandautomatednews.com.