You are on page 1of 4

`CS158-2 Activity #3

Artificial Intelligence creates Electronic Noses

Articles:

How a Computer Chip Can Smell without a Nose


https://newsroom.intel.com/news/how-computer-chip-smell-without-nose/

What is a chemical sensor?


https://www.fierceelectronics.com/electronics/what-a-chemical-sensor

This digital 'nose' could make your coffee smell better and improve your cooking
https://www.zdnet.com/article/this-digital-nose-could-make-your-coffee-smell-better-
and-improve-your-cooking/

Artificial Intelligence Has a Strange New Muse: Our Sense of Smell


https://www.wired.com/story/artificial-intelligence-has-a-strange-new-muse-our-
sense-of-smell/

AI can pick out specific odors from a combination of smells


https://www.newscientist.com/article/2237534-ai-can-pick-out-specific-odours-from-a-
combination-of-smells/

Computers That Smell: Intel’s Neuromorphic Chip Can Sniff Out Hazardous Chemicals
https://newsroom.intel.com/news/computers-smell-intels-neuromorphic-chip-sniff-
hazardous-chemicals/#gs.p1y9ww

'Digital smell' technology could let us transmit odors in online chats


https://www.nbcnews.com/mach/science/digital-smell-technology-could-let-us-
transmit-odors-online-chats-ncna940121

AI is acquiring a sense of smell that can detect illnesses in human breath


https://www.independent.co.uk/news/science/ai-artificial-intelligence-smell-detect-
illness-science-technology-a8394706.html

Instructions:
From above articles, answer the questions below:

This study source was downloaded by 100000767122607 from CourseHero.com on 03-01-2022 08:42:33 GMT -06:00

https://www.coursehero.com/file/109202987/Activity-3docx/
1. Please explain on how an AI can have a sense of smell.

The way in which the architecture of the AI-based olfactory technology is bears great
semblance to how human noses function – specifically how the brain interprets the signals
and stimulus detected by the nose. As described by both the Intel article of March 2020, and
that of ZDNet last September 2020, a “digital nose” incorporates both biochemistry and
modern semiconductor technology to mimic the human sense of smell.
A more detailed explanation of this is that the AI-based technology uses a
combination of biochemical sensors, and optics technology in order to garner data and
detect the required stimulus. This detection of stimulus would then be translated to a unique
set of impulse patterns from the biochemical sensors which would then be interpreted by the
machine through machine learning. By doing so, the machine learns a specified set of
impulse patterns that is quite unique to specific type of scent or gaseous element. The way
this architecture works entails a large number of chemical sensors that would then respond
in the presence of gaseous substances, and a machine learning algorithm that can
categorize what it detects intelligently, mimicking the circuitry of the brain underlying our
olfactory sense.

2. What industries or businesses can benefit from this technology? Please cite three
examples and provide explanation for each industry or business.

The first industry that would benefit with an AI-based olfactory technology is the
safety and security industry. As discussed by an Intel article last March 2020, this technology
offers “smart, reliable, and fast-responding chemosensory systems” that would prove very
beneficial for environmental monitoring and detection of hazardous materials. From
factories, plantations, to transportation hubs like airports, security detection measures will
be more efficient with the application of this technology.
Another industry that would greatly benefit from this technology is the health and
medicine industry. Particularly, this technology can be applied in making accurate diagnoses
for certain diseases that emit a particular odor, in which human sensory detection might not
suffice.
Lastly, businesses within the construction industry would greatly benefit from this
technology. This can be specifically applied to the monitoring of building integrity especially
in the treatment and handling of volatile, hazardous gases and waste that may leak and
cause damage both to properties and lives. By applying this technology, buildings can be
outfitted with detection technology that is more efficient than conventional smoke and gas
detectors that can only detect a limited spectrum of elements.
Apart from this, food and beverage, automotive and even manufacturing industries
will also benefit.

3. What are the possible challenges of e-Noses? (e.g. Distance from subject and AI, you may
use this but provide another challenges)

Possible challenges that e-Noses could and will encounter as further developments
continue are the following: The first one is the proper distinction of similar scents, discussed

This study source was downloaded by 100000767122607 from CourseHero.com on 03-01-2022 08:42:33 GMT -06:00

https://www.coursehero.com/file/109202987/Activity-3docx/
in the intel article published in March 16, 2020. This poses a challenge simply because the
programming architecture of how e-Noses work are direct iterations as to how the human
brain works. Similar as to how neural activity patterns become similar in the presence of
similar odors, the same thing will happen in the olfactory signal recognition of the e-Noses.
Another challenge is the recognition speed. Since e-Noses rely on detecting the
different air molecules to produce distinct signal patterns that would be then interpreted, the
timeframe in which it would recognize a specific scent amid a vast sample size should be
fast enough for accurate findings. This challenge was acknowledged in an article published
in MACH by Parry (2018) in which technology today is limited to heavily relying on air
molecules to hang around long enough to be detected. This reliance of long detection times
also becomes a predicament when a new scent is needed to be determined but a pre-
existing scent lingers, resulting in a muddled and inaccurate results.

4. How close are we now on having an AI that can smell as good as humans?

Given that the most recent articles from March of 2020 states that the progress of
the development of an AI-based olfactory technology is still fresh and a relatively
unventured lands, having an AI that can smell as good as humans would take some more
time. This is primarily due to the challenges aforementioned (technology limitations,
programming architecture, stimulus conditions, environment conditions, etc.).
While Ai-based olfactory technology does exist and work, even applied to certain
products like breath analyzers and modern smoke detectors, an AI that functions how our
noses functions is still a long way to go. This means that real-world applications is still a
ways off (Li, 2020).
To put it into a subjective context, an AI-based olfactory system that functions and
bears a striking resemblance as how the human olfactory sense works, would be in the
process akin to a toddler walking – it works, but it still a lot of gaps for major improvements.
To perceive objectively how close, we are to developing this endeavor is quite difficult for
now, as it is as much an abstract theoretical work as much as it is technology.

5. Do you think that AI will have better smell senses than humans after 25 years?

The question as to whether AI would smell better than humans after 25 years bears
striking resemblance to an age-old question: will AI become more intelligent, perhaps
sentient, to humans? Objectively speaking, the answer to this is no. Unlike pop-fiction and
sci-fi movies that highlight the uprising of the robots, this is quite improbable primarily
because Artificial Technology is being designed today in the likeness of how a human works.

This study source was downloaded by 100000767122607 from CourseHero.com on 03-01-2022 08:42:33 GMT -06:00

https://www.coursehero.com/file/109202987/Activity-3docx/
This means that the development of the AI in any function, even the olfactory, is
limited to how we understand how our system works – dependent on what and how
information is applied in semblance to organic functioning. It is not that AI would “smell
better” than humans do, but it would do it more “efficiently” and how it was supposed to.
The human sense of smell and our understanding of it is limited even now. In an
article by Neuroscience last 2017, a neurologist at Rutgers University in New Jersey named
Jahn McGann, stated in his paper published in Science that human olfactory abilities have
been underestimated and are just as good as other mammals. This means that our sense of
smell is not inferior to that of animals.

Additional Reference:

“Not to be sniffed at: human sense of smell rivals that of dogs, says study.” May, 2017.
Neuroscience. The Guardian. Retrieved from:
https://www.theguardian.com/science/2017/may/11/not-to-be-sniffed-at-human-
sense-of-smell-rivals-that-of-dogs-says-study

This study source was downloaded by 100000767122607 from CourseHero.com on 03-01-2022 08:42:33 GMT -06:00

https://www.coursehero.com/file/109202987/Activity-3docx/
Powered by TCPDF (www.tcpdf.org)

You might also like