Let’s start with a good definition of Artificial Intelligence (AI) “AI is the study of agents that exist in an environment and perceive and act.”(AI A Modern Approach by Peter Norvig)
Conventionally and till now AI is introduced as Agent based system having sensors to observe/ detect/ classify objects from its surrounding/ environment and on the basis of observation our agent/ AI will respond. This concept is correct, every AI system has somewhat same behavior but it’s actions and environment changes or advances in respect of either:
iii) Nature of environment.
Alright, currently whatever we have done in the domain of AI is ANI (Artificial Narrow Intelligence). What does it mean? Let me clear it classically AI started with rule based system like (Chess board, Goal oriented approaches, searching (informed and blind), Shortest path, reasoning, first order logic, and many more). Nowadays we have cool approaches like Machine Learning, Deep Learning, Natural Language Processing, Natural Language Understanding, Reinforcement Learning, Multi-Agent systems etc. I would like to say all these approaches as a tool to achieve ANI.
Note: The aforementioned areas were already introduced in 90 ‘s but now we have discovered brilliant approaches.
Using current power of AI tools/ techniques or approaches what we have achieved?
i) AI beat world best Apha GO player, IBM Watson etc.
ii) AI is taking over customer supports, self driving car, drones etc.
iii) AI is rocking in Healthcare domain and many we even can’t name it.
Alright now it's time to shift our gear towards AGI (Artificial General Intelligence). Starting with definition “ Artificial general intelligence is the intelligence of a machine that could successfully perform any intellectual task that a human being can.” (Wikipedia)
When we talk about AGI there are three category of people/ researchers i) They say AGI will be achieved till 2040 or atleast in this century and after it there will be Utopia, ii) Second class of people/ researchers say it will never be achieved because as human are best and AI can be generalized enough that it may beat human in every part of life (Dystopia), iii) Third class of people say AGI will be achieved but not the way we thinks it like (Utopia or Dystopia), AI has always helped human it's not the replacement of humans. We (humans) will handle AGI as we are handling ANI. Well I’m the person with the third class of people. Why? it’s a great debate if you want to have it then join me with cup of tea (Just Joking!). Back to topic, lets talk about how can we achieve AGI do we have tools that can drive AGI and ASI (Artificial Super Intelligence). So the answer is yes! we have already discovered tools/ approaches that can help us to achieve AGI but we need some integrations, some tweaking of rules, Knowledge. Starting from general intelligence
Here in the mentioned image five components are mentioned to achieve general intelligence. If we want to put general intelligence in machines then we have to integrate the mentioned components. Here I’m listing the areas as a tool which can directly help us to make our pathway towards AI to AGI:
- Learning (Machine learning, Deep Learning, Logical learning etc).
- Logic (Rule base, Knowledge base, Reasoning, Inferencing etc).
- Semantic Web (Knowledge Graphs, Ontologies etc).
How? is a big debate. Last thing about ASI everyone on which everyone is agreed that as soon we achieve AGI we will get soon to ASI.
That’s it for now! I hope you like my story (don’t hesitate to give a applause).
Get connected https://www.linkedin.com/in/siraj-munir-2b59a7b5/
Ping me Siraj Munir
#AI #AGI #MyThoughts