Turing Test, Artificial Intelligence and Human Silliness | Luca Longo | TEDxVicenza
-
0:18 - 0:24In 2016, I was awarded a prize by the
National Forum for Teaching & Learning -
0:24 - 0:28supported and sponsored
by the Ministry of Education in Ireland, -
0:28 - 0:31peculiarly named:
"National Teaching Hero." -
0:32 - 0:34The reason for this award
-
0:34 - 0:37was my availability towards my students
-
0:37 - 0:38and my capability
-
0:38 - 0:43to create less formal and more
comfortable educational environments. -
0:43 - 0:46Let's imagine this place as
a large university classroom. -
0:48 - 0:50I'm used to enter the classroom,
-
0:50 - 0:55[in the] first ten minutes,
when students enter and take place, -
0:55 - 0:59I plug my computer to the speakers
and turn a classical music on. -
1:00 - 1:03I think this is the first step
-
1:03 - 1:07to build less formal and
more comfortable educational environments -
1:07 - 1:10and try to keep the attention
of students at a high level. -
1:12 - 1:14Unfortunately, this is not always easy.
-
1:17 - 1:23In my lessons, I employ a method in use
since the ancient times: storytelling! -
1:24 - 1:26Pedagogically speaking,
-
1:26 - 1:31storytelling is a method
based upon the use of narratives, -
1:31 - 1:33aimed at transmitting
knowledge to students. -
1:34 - 1:38I would start my lesson
exactly with this method, -
1:38 - 1:43by explaining, describing a topic
on everyone's lips nowadays: -
1:44 - 1:46Artificial Intelligence.
-
1:48 - 1:50Like every story narrated to children,
-
1:50 - 1:54I'd like to begin my story with
"Once upon a time." -
1:57 - 2:02Second World War, 1942, United Kingdom,
-
2:03 - 2:04Bletchley Park:
-
2:05 - 2:09a mansion house in the north of London.
-
2:10 - 2:14There was a thirty years old guy:
Alan Turing. -
2:16 - 2:18Alan graduated at King's College,
Cambridge, -
2:19 - 2:21and he obtained
a research doctorate in Logic -
2:21 - 2:24at Princeton University, in the USA.
-
2:25 - 2:30At that time, the Germans made use
of a special machine: Enigma. -
2:32 - 2:34It was like a typewriter:
-
2:34 - 2:36the operator typed some keys,
-
2:36 - 2:41but, instead of printing those letters
on a paper sheet, -
2:41 - 2:43other letters were printed,
-
2:43 - 2:47according to an encoding mechanically
set under the machine. -
2:47 - 2:52The Germans used this machine
to communicate with each other. -
2:53 - 2:56Anybody had listened in on this sheet
-
2:56 - 2:58had in front a meaningless text.
-
2:59 - 3:00It was encrypted!
-
3:02 - 3:08Alan Turing was one of the leading figures
at Bletchley Park: -
3:09 - 3:13he and his team implemented a machine,
-
3:13 - 3:15the one you can see behind me,
-
3:15 - 3:20able to decipher the texts
written by the German. -
3:22 - 3:23Due to this invention we believe
-
3:23 - 3:27that the war terminated two years earlier,
-
3:27 - 3:29saving many human lives.
-
3:30 - 3:34After the war Alan Turing continued
his research in Logic -
3:36 - 3:39and he is considered
the father of Computer Science, -
3:39 - 3:42the father of Artificial Intelligence.
-
3:43 - 3:47With his Turing machine,
he formalises the concept of computer -
3:47 - 3:50even before the computer
was actually built. -
3:52 - 3:57In 1950 he published a paper
on the journal "Mind": -
3:57 - 3:59"Computing Machinery and Intelligence,"
-
4:00 - 4:03where he proposed the Turing test.
-
4:03 - 4:08The question behind the Turing test
is a well-defined one: -
4:09 - 4:11Can machines think?
-
4:14 - 4:16It is at that time that
Artificial Intelligence begun. -
4:19 - 4:22Probably most of you have watched
the movie "The Imitation Game": -
4:22 - 4:24The game of imitation.
-
4:24 - 4:26I am going to describe it to you.
-
4:27 - 4:31Let us suppose that a person is here,
pressing keys on a computer keyboard, -
4:32 - 4:35who asks, pose some questions
-
4:35 - 4:37and on the opposite side of the computer
-
4:37 - 4:40there is a machine M and an operator O.
-
4:42 - 4:44Alternately,
-
4:44 - 4:49the machine M and the operator O
answer to the person P. -
4:50 - 4:55It is said that the machine M
passes the Turing test -
4:55 - 4:58if the person P is not able to understand
-
4:58 - 5:01when the answers come from the machine
-
5:01 - 5:04and when the answers
come from the operator. -
5:04 - 5:06This is The Imitation Game.
-
5:07 - 5:11The machine must have special features
-
5:11 - 5:13in order to pass the Turing test.
-
5:14 - 5:17It has to interpret natural language:
-
5:17 - 5:19the question asked by the person.
-
5:20 - 5:23It has to represent knowledge
in order to formulate answers. -
5:24 - 5:27It has to think in an automatic mode
in order to formulate such answers. -
5:28 - 5:31And it has to learn automatically.
-
5:33 - 5:35There are many approaches
-
5:35 - 5:37to study Artificial Intelligence.
-
5:38 - 5:43One of them is the cognitive approach:
it is based upon the human thinking. -
5:45 - 5:47According to this approach,
-
5:47 - 5:49there are two ways to study
the human thinking: -
5:49 - 5:54either we try to capture thoughts
right when they occur, -
5:54 - 5:57or we try to model thoughts
at a psychological level. -
5:59 - 6:02For this reason,
we say that Artificial Intelligence -
6:02 - 6:06is closely connected to neuronal and
cognitive sciences and to psychology. -
6:06 - 6:09According to this approach,
the assumption is that, -
6:09 - 6:10if we can have
-
6:10 - 6:13a true representation
of the human thought, -
6:13 - 6:16then we can transfer it to a machine.
-
6:18 - 6:21Another approach is the one
based upon the laws of rational thought. -
6:23 - 6:27Probably most of you has heard about
the Aristotle's syllogism. -
6:29 - 6:31Socrates is a man;
-
6:32 - 6:34all men are mortal;
-
6:35 - 6:36[then] Socrates is mortal.
-
6:37 - 6:40This is a deductive reasoning:
-
6:41 - 6:43If we have two truthful premises,
-
6:44 - 6:46we can infer a truthful conclusion.
-
6:46 - 6:48Deductive logics comes from here.
-
6:49 - 6:52According to the laws
of rational thought approach, -
6:52 - 6:57we try to build deductive arguments
and to transfer them to a machine. -
7:00 - 7:03Another approach is
the rational agent approach. -
7:05 - 7:08A rational agent, an entity,
-
7:09 - 7:11has to act,
-
7:11 - 7:13has to adapt itself to the context,
-
7:14 - 7:18has to fix goals to itself,
and be able to carry them out, -
7:18 - 7:21and it has to act in a rational way.
-
7:22 - 7:26Therefore, the Turing test
is related to intelligent agents. -
7:27 - 7:30By rephrasing the Turing statement,
"Can a machine think?", -
7:31 - 7:35we can now say: is it possible to build
a machine, an artificial agent, -
7:36 - 7:38able to think,
-
7:38 - 7:42able to show understanding and rationality?
-
7:44 - 7:46Artificial Intelligence, therefore,
-
7:46 - 7:50aims at developing artificial
intelligent entities. -
7:50 - 7:53Your mobile phone is an entity.
-
7:55 - 7:58By developing artificial entities
-
7:58 - 8:02we try to understand intelligence
as a psychological construct. -
8:03 - 8:05Once we have a better knowledge
-
8:05 - 8:08about this concept,
-
8:08 - 8:11we try to develop
artificial intelligent entities -
8:11 - 8:14to support humans: it's a cycle.
-
8:16 - 8:21But let's see now whether
machines are able to think. -
8:21 - 8:25I'd like to briefly describe you
-
8:25 - 8:28the state of the art
of Artificial Intelligence -
8:28 - 8:30and I'd like to use five classes
-
8:30 - 8:35to classify artificial agents
according to their abilities: -
8:36 - 8:40we have sub-human and par-human agents,
-
8:40 - 8:44over-human, super-human agents
and then we have optimal ones. -
8:44 - 8:46I want to explain them
with some examples. -
8:47 - 8:51Optimal agents are the ones
which act better than all the people -
8:51 - 8:53and you can't do better than that.
-
8:53 - 8:56For instance, agents
which solve the Rubik cube, -
8:56 - 8:59those that play at "Four in a row"
in the best way, -
8:59 - 9:00or at Tic-tac-toe.
-
9:00 - 9:06Consider that some years ago, a boy,
given an initial state of the Rubik cube, -
9:06 - 9:09solved it in 4.73 seconds.
-
9:09 - 9:12Some months ago an agent,
a robot was developed, -
9:12 - 9:15which can solve it in 0.63 seconds.
-
9:17 - 9:22We have super-human agents
which act better than all the humans, -
9:23 - 9:27for instance in the chess game,
in the Scrabble game. -
9:27 - 9:31Some years ago the chess Russian
champion Kasparov -
9:31 - 9:34was defeated by an artificial agent.
-
9:34 - 9:37We have over-human agents,
-
9:37 - 9:40the ones which almost act better
than most of the humans, -
9:40 - 9:45for instance in the Texas hold 'em poker,
in answering the Quiz Show questions. -
9:46 - 9:48We have par-human agents,
-
9:50 - 9:54which act almost like all the humans,
-
9:55 - 9:58for instance in cognitive activities,
-
9:58 - 10:01such as crosswords or image classification.
-
10:02 - 10:06Finally, we have sub-human agents,
which act worse than all the humans. -
10:06 - 10:12Examples include objects classification,
handwriting recognition, -
10:13 - 10:16vocal recognition, translation
from a language into another one. -
10:18 - 10:21But if there is something that artificial
agents nowadays are not able to do -
10:21 - 10:24is for instance, disambiguation:
-
10:24 - 10:29Are we talking about the apple as a fruit,
-
10:29 - 10:32or are we talking about the brand
of the Apple Corporation? -
10:33 - 10:36And one thing that agents
are not able to do -
10:36 - 10:40is reasoning in the real world
under situations of uncertainty. -
10:41 - 10:45These are the main limitations
of Artificial Intelligence, -
10:45 - 10:49and because of these, it is believed that
we are far away to pass the Turing test. -
10:50 - 10:56Now, let's try to understand [whether]
machines will be able to think. -
10:58 - 11:01Some years ago, in America,
a concept has been coined, -
11:01 - 11:03named "Technological Singularity":
-
11:03 - 11:08by Ray Kurzweill, a world-renowned
expert in Artificial Intelligence. -
11:09 - 11:11Let's imagine a timeline.
-
11:12 - 11:17Let's imagine a line indicating
the human intelligence, increasing. -
11:18 - 11:23Let's now imagine a red line
indicating the machine intelligence, -
11:23 - 11:25with an exponential trend.
-
11:26 - 11:31This trend follows the Moore's law,
-
11:31 - 11:35whereby the computational complexity,
for instance, -
11:35 - 11:39as measured by the number of transistors
embedded in a chip, -
11:39 - 11:43doubles every two years
and quadruples every [three] years. -
11:44 - 11:48According to Ray Kurzweill,
in 2010 we should have been able -
11:48 - 11:52to use this computational complexity
-
11:52 - 11:54to emulate the human brain --
-
11:54 - 11:56I didn't see anything.
-
11:56 - 12:00In 2020, with 1,000 dollars
we will have access -
12:00 - 12:03to this computational capacity.
-
12:04 - 12:07In 2025, according to Ray Kurzweill,
-
12:07 - 12:11we will be able to scan our brain
in a very accurate way. -
12:12 - 12:14And eventually, in 2029
-
12:15 - 12:18machines will pass the Turing test!
-
12:20 - 12:24And then, in 2045,
he refers to that point in time -
12:24 - 12:28when the technological singularity
will happen, -
12:29 - 12:33when the machines, machine intelligence,
-
12:33 - 12:35will follow an exponential trend
-
12:35 - 12:39that will significantly affect
the human intelligence. -
12:41 - 12:46In his paper, published
on the journal "Mind," -
12:46 - 12:49Turing not only proposed his test,
-
12:49 - 12:53but he also suggested nine objections
against his own test. -
12:53 - 12:58These objections are nine objections
against Artificial Intelligence. -
13:00 - 13:04Some years ago, when I was a student
at the University of Varese, -
13:05 - 13:09I attended a course on "Epistemology,
Deontology and Ethics in Computer Science" -
13:09 - 13:13held by Prof. Gaetano Aurelio Lanzarone.
-
13:13 - 13:16Unfortunately, he passed away
some years ago. -
13:16 - 13:22One of the assignments we had to do
was to propose a tenth objection -
13:23 - 13:26against the Turing test,
against Artificial Intelligence. -
13:27 - 13:30I was the only one who proposed
a tenth objection -
13:31 - 13:33expressed as a mathematical equation,
-
13:35 - 13:37that I labelled "human stupidity."
-
13:37 - 13:40I'd like to explain it in simple terms.
-
13:40 - 13:44Let's assume we take
the intelligence of all humans -
13:44 - 13:46and we put it all together.
-
13:46 - 13:49The Sum symbol of the equation
on the left hand side. -
13:50 - 13:53And we transfer this intelligence
as a whole to a machine. -
13:54 - 13:58Then, we get an equality of intelligence.
-
13:58 - 14:01But in some way the machine
becomes more intelligent than us. -
14:02 - 14:04Though, if it is true that were us
-
14:05 - 14:09who have transferred our intelligence
to a machine -
14:09 - 14:12and it becomes more intelligent than us,
-
14:12 - 14:13it is also true, as well,
-
14:13 - 14:17that we let the machine become
more intelligent than us. -
14:19 - 14:22Then, in order to conclude my story,
-
14:23 - 14:28and referring back to the initial
question: "Can machines think?", -
14:30 - 14:34I'd like to leave you with
an open question: -
14:36 - 14:40Does really makes sense
for us to let them think? -
14:41 - 14:42Thank you.
-
14:42 - 14:44(Applause)
- Title:
- Turing Test, Artificial Intelligence and Human Silliness | Luca Longo | TEDxVicenza
- Description:
-
Luca Longo is currently assistant professor at the Dublin Institute of Technology, where he is a member of the Applied Intelligence Research Centre. His core research interest is in Artificial Intelligence, particularly in Mental Workload modelling. His talk starts with the question: "Can machines think?" and it ends up asking himself: "Does really makes sense for us to let them think?"
This talk was given at a TEDx event using the TED conference format but independently organized by a local community. Learn more at https://www.ted.com/tedx
- Video Language:
- Italian
- Team:
- closed TED
- Project:
- TEDxTalks
- Duration:
- 14:47