===========================================================================

IDENTIFYING INFORMATION:

Title: Is Artificial Intelligence Always Intelligent?

Reference: BSubm19



1) Type of paper

a) _____ Research paper (with original results)

b) __X__ Application paper (experimentation __X__ / case study ____ )

c) _____ Survey of recent advances

d) _____ Position paper

e) _____ Other. Please specify ______________





2) General evaluation

Please rate the following criteria (a,...,f) by, each time, using only one

of the five following words: BAD, WEAK, FAIR, GOOD, EXCELLENT



a) Relevance to NICSO: FAIR

b) Originality: WEAK

c) Significance; usefulness: BAD

d) Technical soundness: BAD

e) Reference to the related literature: WEAK

f) Presentation: BAD





3) Summarized description of the paper (Please provide brief answers.)





- What is (are) the main contribution(s) of the paper to the field?

The paper tries to state a question if intelligent behaviour could be called
intelligent if it lacks ability to adapt to the changing environment. A neural
network is used to train an agent travaling amongst obstacles on a 2D grid.
A trivial 2D extension of a 1D jeep problem is proposed. However none of
this can be considered as a significant contibution.

- How important is this contribution w.r.t. the state of the art?

Although the initial idea of assesing if the machine learning techniques like
neural networks are really able to generate intelligent and adaptive behaviour
is very interesting and releveant the paper does not answer to that question
at all.


4) Discussion of technical aspects (detailed comments)

Please answer the following questions to any level of detail you find
suitable :



- Indicate difficulties, flaws which weaken the analysis of the problems and
the claims.

The observation that the behaviour of an agent does not change under unchanged
conditions for the neural network with random weights is trivial. Only the use
of a learning mechanism (e.g. reinforcement learning through back propagation)
could change the agents behaviour in consecutive runs. Although the authors
introduced a mechanism of weights altering during simulation, their observation
that behaviour (route) is different for different set of initial NN weights is
again trivial. The authors also ignored the fact that when obstacles have been
introduces the agent did not learn how to avoid them is a consequence of how
they designed the learning mechanism and do not elaborate on that issue at all.
The extension of the jeep problem was introduced without any resoning on how
its benchmarking power is different from the previous experiments or the original
1D problem. Also no results using the proposed benchmark were presented in the
paper.

- In what respect does the paper lack originality, and/or what are the
missing references?

As authors stated themselves in the conclusions, the work presented here is
at the level of the undergraduate student project. It lacks originality, it
lacks literature review, it lacks organisation and deeper thought of what is
the goal of the research and how to get there.

- Technical soundness

a) ___ Technically correct (after careful check)

b) ___ Apparently correct (after superficial check)

c) ___ Minor errors (please indicate them):

d) _X_ Major errors (please indicate them):

- the design of experiments does not support the testing of the hypothesis
- the number of experiments performed is not stated
- the measure of success is not clearly defined
- the analysis of results is very superficial

e) ___ Unsupported claims (please provide a detailed explanation):



- Linguistic quality (Indicate English mistakes to be corrected, or send us

back the manuscript with proper corrections.)

a) ___ Very good

b) ___ Acceptable

c) _X_ Unreadable


Many sentences are grammatically incorrect. The colloquialisms and
the informal style is overused.



********************************************

GENERAL RECOMMENDATION (choose only one)

a) ___ strong accept

b) ___ accept

c) ___ minor revision

d) ___ major revision

e) _X_ reject



Main reasons for your decision:



- For choices a, b please indicate:

1) ___ accept because of the originality (good ideas, sound presentation)

2) ___ accept because of the quality of the proposed synthesis (useful

review on recent advances)



- For choices c, d please indicate what has to be revised:







- For choice e please indicate:

1) ___ reject because it is not relevant 

2) _X_ reject because of the presentation

3) _X_ reject because the content is too premature

4) _X_ reject because of lack of originality

(results already known; similar overview already published)

5) _X_ reject because of major errors



--------------------------------------------------------------

*** REMARKS FOR THE PROGRAMME COMMITTEE:

If you wish to add any remarks for PC members, please write them below.

These remarks will only be used during the PC meeting.

They will not be sent to the authors.

This field is optional.
===========================================================================