Page 94 - วารสารกฎหมาย ศาลอุทธรณ์คดีชํานัญพิเศษ

วารสารกฎหมาย ศาลอุทธรณ์คดีชำานัญพิเศษ

office again and this time he was told that winter is going to be the coldest on
record. The chief being part concerned and curious then asked, ‘How can you

be so sure?’ to which the meteorologist replied, ‘The local Native Americans
are collecting wood like crazy’.

[Emphasis added]

On 23 March 2016, Microsoft released Tay, an AI chatterbot that was supposed

to mimic a 19-year old American girl learning and interacting with human users of
Twitter. However, within 16 hours of its release, Tay had to be taken down as it started
posting offensive and inflammatory tweets instead of fulfilling its intended purposes as

a scientific attempt to create the AI of an innocent and conversational bot eager to learn
from human. According to a statement by Microsoft, Tay’s horrendous behaviour was
blamed on a ‘coordinated attack by a subset of people exploited a vulnerability in Tay’

and it would appear that this was the result of an exploitation of the chatbot’s ‘repeat
after me’ function, in other words, Tay’s misbehaviour was consequent to it been taught
19
all the wrong things by humans. An algorithm is a reflection of what we want it to be,
as well as the data we feed it with. According to some AI thinkers and researchers,
many of the key algorithms that affect public life are also considered proprietary or

trade secrets. Consequently, ‘such opaque, uninformed understanding of algorithms
impedes intelligent public discourse on their shortcomings.’ Must algorithms be open
20
to public scrutiny, validation or verification? If so, how will it be regulated, by who and

according to what standard?

It is also worthwhile considering the case of Catalina v Norma involving
21
a Portuguese owned vessel, Catalina. The sole arbitrator was heard by the parties saying
that ‘Italians are liars and so as the Portuguese’ whereas the other party being Norwegians

19 Abby Ohlheiser, “Trolls Turned Tay, Microsoft’s Fun Millennial AI Bot, into a Genocidal Maniac’
Washington Post (Washington, 25 March 2016) <https://www.washingtonpost.com/news/the-intersect/wp/2016/03/24/
the-internet-turned-tay-microsofts-fun-millennial-ai-bot-into-a-genocidal-maniac/> accessed 23 March 2021
20 Osonde Osoba, ‘An Intelligence in Our Image: The Risks of Bias and Errors in Artificial Intelligence’
(n 10) 6.
21 61 Lloyd’s Law Reports 360 et seq (1938)

92