Artificial Intelligence
Technology is very common words in our daily life’s.
Everything which man has invented after Stone-Age is a part of technology. Technology
can be the knowledge of techniques, processes, etc. or it can be embedded in
machines, computers, devices and
factories, which can be operated by individuals without detailed knowledge of
the workings of such things.
Common
philosophy related with AI?
1.
Common philosophy related to AI
for general public
a)
Some basic AI Example (SIRI,
Google Now/ Search)
Newspapers, magazines, the Internet in
general have been throwing out the terms “AI” and “Artificial Intelligence” a
lot.
From
SIRI to self-driving cars, artificial intelligence (AI) is progressing rapidly.
AI can encompass anything from Google’s
search algorithms to IBM’s Watson to autonomous weapons.
WEAK
AI & Strong AI (comparison)
1. Weak AI- is designed to perform a narrow task
(e.g. only facial recognition or only internet
searches or only driving a car).
a)
Examples
only facial recognition or only internet searches or
only driving a car
b)
Benefits
While narrow AI may outperform humans at whatever its
specific task is, like playing chess or solving equations,
2.
Strong AI
a)
Example- Doing anything
b)
AGI would outperform humans at
nearly every cognitive task.
However, the long-term goal of many
researchers is to create general AI (AGI or strong AI).
A lot of people assume that we are
developing general AI rather than applied AI.
What actually aI
is?
Things to achieve "learning" and
"problem solving".
The central
problems (or goals) of AI research include reasoning, knowledge, planning,
learning, natural language processing (communication), perception and the
ability to move and manipulate objects. General intelligence is among the
field's long-term goals. Approaches include statistical methods, computational
intelligence, soft computing (e.g. machine learning), and traditional symbolic
AI.
1.
Artificial Intelligence è Machine Learning è significantly Deep Learning
Artificial Intelligence is not artificial
intelligence. When people talk about ‘volcanic’ changes in ‘AI’ they are
talking about one particular field of technology: Machine Learning (most
significantly Deep Learning). Machine Learning is a very literal description of
the technology it describes, that is a program written to learn and adapt.
How is artificial intelligence programmed?
1.
Analogy
If anyone claims to know
the answer to this question, then ask them to show you their working AI
program. If they claim to have a program that could be trained to do any job
that a human could do, then ask them why we are still paying people worldwide
USD $75 trillion per year to do that work instead.
2.
Real Answer
Nobody knows how AI is
programmed.
3.
How to someone work in AI-
It's not that we aren't
making progress. Google and Facebook have invested heavily in AI. Google search
does a pretty good job of understanding what you want when you speak or type a
question in natural language. Facebook is pretty good at recognizing faces and
understanding your interests. The reason these companies are making progress is
that they have thousands of developers, millions of CPUs, Exabyte of data, and
hundreds of billions of dollars.
4.
Analogy to human mind
AI is an extremely hard
problem. But we can at least gain some insight into the hardware and software
requirements of AI by examining the only known working example of human
intelligence.
The human brain has about
10^11 neurons and 10^14 synapses. An artificial neural network of this size
with response times of 10 to 100 milliseconds would require 10^15 to 10^16
operations per second. Such computers exist, but they cost millions of dollars
and consume several megawatts of power.
Doing the work of humans
requires both a mind and a body. Therefore, we should expect that an AI
controlled robot that could be trained to do any task that a human could do
should be as complex as a human. Complexity is measured in bits: it is the size
of the software when compressed using the best possible algorithm.
I actually did this
experiment, where I compressed the human genome (our source code) and compared
it with a large collection of open source code. The answer is that a human baby
has the same complexity as 300 million lines of code (after which it will still
require years of training). Software costs about $100 per line. This makes the
total cost $30 billion. This is why only large companies are making significant
progress in AI.
How close are big companies in
developing AI
1.
One AI system will train
another one.
We are getting closer to general
AI though. There is a developing technology, “Adversarial Training in Recurrent
Neural Networks “, where the data from one machine learning program helps to
train the other. This is the technology that Google and Facebook have been
flouting a lot recently. An example of this might be in medicine, where one ML
program is used to diagnose a patient, and another is used to prescribe a
treatment. The two programs may train each other in that correct treatments
suggest correct diagnoses and the correct diagnosis may lead to different
treatments, and so on…
2.
How AI will help us and case
example.
There are companies that
have already hinted they are riding the AI/ML wave. Travis Kalanik has
suggested on multiple occasions Uber has no need for drivers. Instead,
driverless cars powered by ML technologies could hack the way for the future of
transportation. Why is a taxi company worth over USD 45 billion? Because it
isn’t about taxis, it is about automating and optimizing global infrastructure
in general. That is an exciting vision: powered by AI. That is also going to
kill multiple industries, not merely hasten their decline.
HOW
CAN AI BE DANGEROUS?
Most researchers agree that a super
intelligent AI is unlikely to exhibit human emotions like love or hate, and
that there is no reason to expect AI to become intentionally benevolent or
malevolent. Instead, when considering how AI might become a risk, experts think
two scenarios most likely:
The
AI is programmed to do something devastating: Autonomous weapons are artificial
intelligence systems that are programmed to kill. In the hands of the wrong
person, these weapons could easily cause mass casualties. Moreover, an AI arms
race could inadvertently lead to an AI war that also results in mass
casualties. To avoid being thwarted by the enemy, these weapons would be
designed to be extremely difficult to simply “turn off,” so humans could
plausibly lose control of such a situation. This risk is one that’s present
even with narrow AI, but grows as levels of AI intelligence and autonomy
increase.
The AI is programmed to do something
beneficial, but it develops a destructive method for achieving its goal: This
can happen whenever we fail to fully align the AI’s goals with ours, which is
strikingly difficult. If you ask an obedient intelligent car to take you to the
airport as fast as possible, it might get you there chased by helicopters and
covered in vomit, doing not what you wanted but literally what you asked for.
If a super intelligent system is tasked with an ambitious geoengineering
project, it might wreak havoc with our ecosystem as a side effect, and view
human attempts to stop it as a threat to be met.
As these examples illustrate, the concern
about advanced AI isn’t malevolence but competence. A super-intelligent AI will
be extremely good at accomplishing its goals, and if those goals aren’t aligned
with ours, we have a problem. You’re probably not an evil ant-hater who steps
on ants out of malice, but if you’re in charge of a hydroelectric green energy
project and there’s an anthill in the region to be flooded, too bad for the
ants. A key goal of AI safety research is to never place humanity in the
position of those ants.
Future with AI
Before you dismiss AI as decades away, keep
in mind that our expectations of development in technology are linear whilst
its progress is exponential. Today, we may be reading with sci-fi goggles, the
progress of Machine Learning, but by the time our cars drive themselves the
cities we drive them in will radically change. Neither will cars look like cars
anymore.
In the meantime, IMHO it is worth
monitoring the impact of Machine Learning on these industries:
Education
Agriculture
Medicine
The Military
Infrastructure
Transportation
References
Comments
Post a Comment