It gets harder every year to find a tech article without some reference to the word AI ( Artificial Intelligence). There is no getting around the fact that AI is the next big thing. While many well-meaning thought leaders have issued warnings regarding the dangers of AI. These dangers are not anywhere in sight. Not yet anyway. Well, whatever be the case there is no stopping science.
Let’s have a look at 10 ongoing AI projects that will blow your mind away and make you question about what machines can do.
1) Project: OpenCog
OpenCog is an open-source software project with the aim to build a human-level artificial general intelligence (AGI). Despite the objective of human level, OpenCog doesn’t emulate the human brain in any discernable detail. Instead, it tries to use the currently available computer hardware to run custom software that draws its inspiration from neuroscience, cognitive psychology, and of course, computer science. The underlying assumption is that the human brain is only one particular way of achieving general intelligence and that some other method can be just as viable. Ben Goertzel leads the project. The roadmap target is reaching human-level AGI by the end of 2021. OpenCog is being used in various practical projects such as video games, info retrieval, biomedical informatics (analyzing DNA data, genetic analysis) hedge fund / financial predictions on the Hong Kong stock exchange.
2) Procedural Reasoning System (PRS)
In artificial intelligence, a procedural reasoning system (PRS) is a framework for creating a real-time reasoning system that can perform complex tasks in dynamic environments. It is based on the idea of a rational agent using the software model based on belief-desire–intention.
A user application is usually defined, and provided to a PRS system is a set containing knowledge areas. Each knowledge area is the procedural knowledge that specifies how to do a particular task, e.g., how to navigate a corridor etc. Such programs, together with a PRS interpreter, control the agent. The interpreter is responsible for maintaining the beliefs regarding the world, choosing which goals can be achieved next, and choosing the correct knowledge area to apply in the given situation. Unlike traditional AI planning systems that have a complete plan at the beginning, and replans the whole things if something unexpected thing happens, PRS interleaves planning and doing actions in the world. PRS is based on the BDI or belief–desire–intention framework. Beliefs are what the agent believes to be correct about the current state of the world, desires are the agent’s goals, and intentions are the agent’s current plans for achieving the goals.
Also Read: How Machine Learning is changing the world
3) CALO ( Cognitive Assistance that learns and organizes)
It was an artificial intelligence project that attempted to integrate various AI technologies into a cognitive assistant. The project started in May of 2003. The project ran for five years ending in 2008. The CALO project had significant spin-offs, most notable being Siri intelligent software assistant. It is now a part of the Apple iOS since iOS 5.
Defense Advanced Research Projects Agency (DARPA) funded CALO under the Personalized Assistant that Learns (PAL) program. The five-year-long project brought together about 300 researchers from top university and commercial research institutions, with the goal of building cognitive assistants that can reason, learn from experience, and respond stably to surprise. SRI International was the lead integrator for the project. They passed a yearly evaluation that measures how well the system has learned to do its job.
MYCIN is a project that worked a little early than the rest of the projects on the list. It was a backward chaining expert system that employed artificial intelligence to identify the bacteria causing infections, such as bacteremia and meningitis, and recommending antibiotics based on the patient’s body weight. The project was the dissertation of Edward Shortliffe under the direction of Bruce G. Buchanan, Stanley N. Cohen and others. MYCIN received an acceptability rating of 65% on a treatment plan from a panel of eight independent specialists. MYCIN was never used in practice anywhere. This wasn’t because of its weak performance but due to the ethical and legal issues raised related to the use of computers in medicine. But, the most significant problem was the state of technologies especially at the time it was developed.
Also Read: 10 Security Tips for safe Internet browsing
6) Google Assistant
Perhaps the most famous and commercially available. Google Assistant is a virtual assistant available on mobile phones and smart home devices. Unlike Google Now it’s predecessor, the Assistant can engage in two-way conversations with the user. The User interacts with the Google Assistant through natural voice, though keyboard support is also included just in case. In a similar fashion to Google Now, the Assistant can search the net, schedule events and adjust hardware settings on the user’s device. Recently Google has announced that the Assistant will be able to identify objects and gather visual data through the device’s camera, as well as identification of songs.
Psi-theory is a psychological theory covering human action regulation, intention selection and emotions. Dietrich Dörner developed it at the University of Bamberg. It models the mind of a human being as an information processing medium, controlled by a collection of physiological, social and cognitive drives. Perceptual processing is modulated by these drives, which allow for autonomous establishment and pursuit of goals in an open environment. Next, to the motivational and emotional system, Psi-theory also suggests a neuro-symbolic model of representation, which encodes semantic relationships in a hierarchical spreading activation network. Psi-theory has many applications. Open cog includes an implementation of Psi-theory. MicroPsi is a cognitive architecture based on psi-theory. Also Advanced Magic uses psi-theory for natural language processing in artificial intelligence applications.
DUAL is a general cognitive architecture combining the connectionist and symbolic approaches at the micro level. This architecture is based on decentralised representation and emergent computation. It was inspired by the Society of Mind idea proposed by Marvin Minsky, but it departs from the initial proposal in various ways. Calculations in DUAL occur from the interaction of many micro-agents. each micro-agent is a hybrid symbolic/connectionist device. The agents exchange messages via links that can be modified. They form combinations which collectively represent concepts, episodes. Several other models have been developed with the help of DUAL. These include AMBR, PEAN (a model of perception), JUDGEMAP (a model of judgment) etc. A Team led by Boicho Kokinov at the New Bulgarian University developed DUAL. Alexander Petrov co-authored the second version.
AIXI is a mathematical formalism for artificial general intelligence. It combines the Solomonoff induction with the sequential decision theory. Marcus Hutter first proposed the formalism in 2000.
It is a reinforcement learning agent; it maximises the expected total rewards from the environment. Intuitively, it concurrently considers every computable hypothesis. At each time step, it looks at every possible program and evaluates how many rewards that program generates depending on the action taken. Then the promised rewards are weighted by the subjective belief that this particular program constitutes the real environment. This subjective belief is computed from the length of the program: longer the program, less likely its possibility. Then AIXI selects the action that has the highest expected total reward in the weighted sum of all the programs.
9) LIDA (Learning Intelligent Distribution Agent)
The LIDA cognitive architecture is an artificial cognitive system that is an attempt to model a wide spectrum of cognition in biological systems, from low-level action to high-level reasoning. This model was primarily by developed Stan Franklin and colleagues at the University of Memphis. The LIDA architecture is grounded in cognitive science and cognitive neuroscience. Giving working explanations for many cognitive processes, the LIDA conceptual model is also sometimes intended as the way with which to think about how minds work. There is 2 hypothesis at the heart of the LIDA architecture:
1) Most of the human cognition functions employing frequently iterated (~10 Hz) interactions, also known as cognitive cycles, between conscious contents, the different memory systems and the action selection.
2) These cognitive cycles, serve as the “atoms” or “core” of cognition which compose the higher-level cognitive processes
10) IBM WATSON
It is a question-answering computer capable of answering questions posed in natural language, developed by a research team led by principal investigator David Ferrucci under IBM’s DeepQA project Watson is named after IBM’s first CEO, Thomas J. Watson. At first, the system was developed to answer questions on the quiz show Jeopardy. In the year 2011, Watson competed on Jeopardy! Against legendary champions, Brad Rutter and Ken Jennings and won the first place prize of $1 million.
In February 2013, IBM announced Watson system’s first commercial application. It would be for utilization management decisions in the treatment of lung cancer at Memorial Sloan Kettering Cancer Center, New York City. This was done in conjunction with health insurance company WellPoint. Manoj Saxena, IBM Watson’s former business chief, says that 90% of nurses who use Watson now follow its guidance.
Top 10 ongoing AI Projects that will blow your Mind
I am a sophomore at IIT Roorkee currently pursuing my B.tech in Metallurgical and Materials Engineering.