Contributing researcher: Aparajita Bandopadhyay

Languages in Translation: Wasn't one language enough?

To quote Franz Kafka, who, interestingly enough, was a brilliant renegade of a writer- ‘All language is but a poor translation’.

• Many foreign words, which have resisted easy translation, have simply been incorporated without any change into English- e.g., schadenfreude and weltschmerz- and some linguists, such as Bickerton, have developed a theory which roughly states that a people which does not experience certain events,or know of certain objects, will not create words corresponding to them. This is an intuitively obvious idea; but, quoting Matt Ridley, it would be absurd to argue that only Germans can understand the concept of schadenfreude, and the rest of us find the concept of taking pleasure from the sorrows of our neighbours foreign.
• Redundancies in words, too, inevitably appear in the development of languages. We must also call into consideration the fact that omnipresent and abundant things cannot, logically, have words corresponding to themselves; why would we want a word for something that is taken for granted? After all, the things we know the best are the things we don’t know that we know. Even this may serve as a hindrance to our complete comprehension of another language. It is also a highly engaging activity for our brains to learn and communicate with languages. Our brains use our Broca’s areas, Wernicke’s areas, prefrontal cortices, and several other centres for constructing simple sentences.
• The fact that languages have influenced us greatly is proved by most of our inabilities to instantly grasp a new language- yet it is ourselves who have influenced languages in such diverse ways that different ones are hard to grasp. Languages are yet another proof of the often laughable circularity underlying human thought (every definition is ultimately circular-for example- the definition of a ’thing’ can be an ‘object’, and that of an ‘object’ may be a ’thing’-upon being asked what both of them mean, we can smugly reply-’the same thing as each other!’).Our inevitable ineptitude for the expression of our most elementary axioms reveals not only the visceral and nebulous nature of our concepts, but also the idea of a strange, uniform camaraderie between us, as a race. 
• Languages delude us about ease of communication very often; yet, we often laugh at our own complexity in thought when trying to straighten out paradoxes of clever phrasing,and oxymorons (how many of us have really heard silence?).

No wonder then, that the best of Google Translator APIs and NLP/ language translator modules from Microsoft Azure or AWS, often generate translated text-strings that read more like AI-generated lame jokes.
One good service they all provide to human intelligence though- they augment our sense of humour- "Maria...makes me... laugh"!- so goes the famed song from The Sound of Music! 

Contributing researcher: Aparajita Bandopadhyay

As defined by Robert Sapolsky in his lecture on ‘Language’ at Stanford, human language cannot be considered 'language', without these seven key features that distinguish it from the communication between other creatures. 

These include:

1. semanticity, 
2. embedded clauses, 
3. recursion, 
4. displacement, 
5. arbitrariness, "Motherese", 
6. meta communication, and 
7. prosody.

Each of these facets of human language are exclusive to our species, and one or the other is typically the reason that various apes who were taught ASL (American Sign Language) over the years, have been unable to recreate language that is truly ‘human’. 

For communication to be complete, practical and real in the human world, each of these features is essential. In the same way, AI with Natural Language Processing/ Comprehension & Understanding/ Generation of contextually & semantically sensible dialogs, could only become humane if these features were embedded into the famed Siris and Alexas. The difficulty lies in creating these features without compromising datasets or precision, thereby reaching the holy grail of Artificial General Intelligence (AGI). 

1) Semanticity- the ability to generate and convey meaning, by ‘bucketing’ sounds to create words, is the most fundamental feature of any and every human language. The instinctiveness of meaning makes it all the more difficult to completely transfer meaning into a system, and clearly makes it incredibly challenging to generate meaning in novel words and communicate successfully. The awareness of semanticity in itself, the definition of semanticity that thereby gives the word meaning, is an almost inexplicable, and simultaneously intuitive, concept.

2) Embedded clauses are arguably the easiest to represent through programming languages as well as human languages, using simple to complex logic constructs- from Aristotle's term logic to propositional and predicate logic that form the foundations of AI. 

• Adding conditions or details to a pre-existing clause, that answer questions such as where, when, how or why, in relation to the clause, creates language rich in information. 
• Specifications through embedded clauses result in more informative sentences, and thus more meaningful communication. 
• If further detailing and longer sentences do provide more information in a sophisticated algorithm, this facet can be conquered.

3) Recursion or generativity is an incredibly interesting property of language: a finite number of words can produce an infinite number of sentences, and a sentence can have infinite length, bounded only by practical time constraints. 

• By embedding a sentence in another sentence with perhaps an identical format, you can generate a longer, different sentence. ‘They don’t know that we know they know we know’, is a perfect example that could technically last forever continued in the same exact pattern, without losing meaning. 
• Recursion in programming usually refers to self-referencing, such as recursive functions that call themselves within the function definition.
• Generativity is relatively easy for people to manufacture, by adding clauses, and may be easy to manufacture randomly as well, with strict boundaries of meaningfulness.

4) Displacement is the one feature that most famous chimpanzees and gorillas learning ASL could not achieve. Displacement involves the ability to talk about different time periods, different people, regardless of present circumstances, and without only conveying current emotion. Being able to talk about things emotionally distant from us, and dissociating our communication from our situation, is a unique capability that we seem to convey easily. Giving each other information in ‘facts’ not directly pertaining to ourselves, or even getting the answer from a chatbot when you ask for 23rd January’s weather, are examples of displacement ingrained in human language.

5) Arbitrariness refers to the lack of connection between the meaning of words and their shapes or sounds. Adjectives such as ‘heavy’ or ‘sad’ are not created with relation to the shapes of the letters in them, or resembling the sounds they make. The arbitrariness of human language makes it difficult to compare languages by sound or script, since neither of the two convey meaning. The randomness of meaning is, therefore, immune to guesswork or brute force.

6) Meta communication, the ability to communicate about communication and discuss language at all, is another feature of human language, forming the basis for natural language processing and generation. Meta communication also refers to secondary communication that can change or add to the meaning of a sentence or communication. 

7) Prosody, which forms a derivative part of meta communication, involves things such as intonation, stress, rhythm and other parts of body language that accompanies a unit of communication. The accompanying body language and intonation can convey different meanings to even the same sentence, which can be expressed through meta communication, such as sarcasm created using varied tones of voice.

Motherese, or baby talk, consists of the difference in intonation, including stress on vowels, that parents use when communicating with their infants, with the intention of teaching them how to speak. Specific to humans, it is an important part of child language acquisition, a field growing in importance to machine learning through natural language acquisition.

Each of the 7 features that distinguish human language from others seems instinctive, and unnoticeable in daily conversation. To train a language model on these basic but generic features of human languages, is a different ballgame altogether.  

Ian Head

Contributing researcher: Rudrani Bose

Since 2019, most companies have been using AI to improve process performance and costs, and some of the early mover multi-country organizations are also focussing on leveraging AI to improve customer experience. They are doing it by using AI to assure uniform levels of consistency, reliability, performance and service quality, across their regions of operations. 

Now Ethics in AI has become a key issue especially when companies are using AI and intelligent automation across their value spectrum and supply chains & partner ecosystems. AI developers must ensure that their deployments are able to comply with the ethical requirements in all the regions where their systems will be deployed. Breaking local privacy laws is invariably seen as arrogant and intrusive and malicious intent is often inferred. 

In this context, ethical use of AI in the enterprise becomes the most critical change pivot. AI-user (employee/ customer) experience is significantly influenced by the perceived ethical behaviour of the AI solutions. This is applicable for all enterprise AI solutions regardless of specific domains and usecases, e.g. conversational AI agents (e.g. customer service chatbots) or digital workers (e.g. intelligent ops process bots for claims processing, invoice processing, KYC, duplicate invoice detection, false claims & fraud detection, retail shrinkage detection, and so on). 

All global enterprises have to work within the government policy frameworks in their regions of operations, and strategic AI usecases must often comply with multiple local government policies and priorities on ethical AI. International and global use cases require policy adaptability so as to adhere to multiple regional idiosyncracies; as US companies working in the European Union have discovered (1).  

Here are some examples of current  government key initiatives on ethical AI. 

The Australian government released a report along with CSIRO listing out the framework for an ethical AI. This states that any AI system must:

1. Control any adverse or unethical effects of AI to ensure that the generation of net benefits for citizens are greater than the costs and risks of adopting and developing AI and associated technology stacks

2. Demonstrate regulatory and legal compliance with international and governmental guidelines to prevent AI systems from harming or deceiving citizens 

3. Ensure the protection of citizens’ private data and prevent data breaches causing financial, psychological or professional harm

4. Develop controls to monitor the fairness and inclusivity of data systems and restrain damage from biased algorithms. The term “bias” has expanded scope considerably in recent years. 

5. Ensure adequate transparency, responsibility and accountability of governments, organisations and people by ensuring that identifiable data is only used after gaining the consent of each citizen.

The Treasury Board Secretariat (TBS) of Canada is working towards the responsible use of AI in government. It is working with the international AI community to facilitate an open online consultation to foster collaboration on international guidelines for the ethical use of data systems. 

The federal government wishes to lead by example through adopting healthy ethical practices in its use of artificial intelligence. To ensure the proper and ethical use of AI, the government plans to:

1. Estimate the impact of using through the development and      collaboration on standard tools and approaches
2. Adopt transparent and accountable practices in AI decision-making      and carefully review the use of AI, along with the user benefit generated.      
3. Enable openness in data sharing while ensuring national security,      defence and integrity of personal and organisational data and information.      
4. Provide adequate training to upskill government employees in      possessing the capability to facilitate the development, design and usage      of reliable AI solutions and public services. 

The American AI Initiative launched by President Donald Trump labels itself “AI with American Values” in seeking to protect and preserve freedom, human rights, institutional stability, the rule of law, right to privacy, intellectual property rights and equitable opportunities for all.  Right to privacy itself is not stated and defined clearly in the constitution, Unlike the EU where right to privacy is clear and mandatory e.g. by GDPR. 

In keeping with that philosophy, in February 2020, the US Department of Defense adopted a guideline of ethical principles to govern the use of AI. The regulations seek to incorporate responsibility, equitability, reliability, traceability and governability in digital technology used not only for military and defence purposes but also for commercial and for-profit business objectives. It aims to maintain the strategic global leadership of the US and respect the rules-based international order.
However US corporations have encountered difficulties when they assume that compliance with US law confers compliance with laws in other regions. For example in the European Union, where the right to privacy is strictly stated in the General Data Protection Regulations (GDPR).

The government of the United Kingdom defined the ethical framework required to design responsible digital systems as those which 

- respect the dignity of individuals and allow people to form open, meaningful and inclusive connections. 

- Protect social values, social justice and prioritise public interest by providing for the wellbeing of all.

The Centre for Data Ethics and Innovation was set up by the government in 2018 to help navigate the ethical challenges posed by rapidly evolving digital systems. For this purpose, the Alan Turing Institute, the Office for Artificial Intelligence and the AI Councilwere also set up. As the UK leaves the EU, it is expected that these bodies will publish new recommendations to replace and possibly enhance the EUs General Data Protection Regulations (GDPR)

Compared to other countries, Singapore has yet to develop a concrete strategy to incorporate ethical practices in AI systems. In June 2018, Singapore announced the establishment of an AI ethics advisory council  on the development and deployment of artificial intelligence, to guide the government in building ethical standards and codes of conduct for businesses. 

We can expect this body to produce guidance or regulations in the near future but whether the requirements will be similar to those of other bodies referred to in this document, is as yet unknown. 

(1) Facebook warns it may be forced to pull out of Europe  https://www.telegraph.co.uk/technology/2020/09/21/facebook-warns-could-pull-services-europe/ 

(2) https://www.theleaflet.in/specialissues/right-to-privacy-in-the-united-states-of-america-by-nehmat-kaur/

(3) https://thehill.com/opinion/judiciary/445975-what-do-you-mean-theres-no-right-to-privacy-in-america 

 Finally, intelligence of all kinds- artificial or human - has prevailed. The new President-elect for the United States, speaks for Science, Rationality, Fairness. A huge breather for the IT and tech companies, in product/ platform/ services and adjacent domains.

Being the Largest economy of the World, and also as home to some of the world’s best academic and research institutions, social and policy stability in the US does influence the Quality of Life and access to Knowledge & tech, for the Whole World.

Interesting shifts are already seen in the Tech world- Top tech companies like Microsoft, Google and IBM, have CEOs of Indian origin. Now Kamala Harris brings fairness in political leadership too. Balancing out the over-Indian flavour in these curries, Interesting pivots are seen in some of the Big Indian ITSPs switching to non-Indian leadership at CEO levels.

Fantastic amalgamation of East-meets-West cultures n value systems. Another interesting switch in this fact is that, now, some top tech Products n Platforms co.s have Indian leaders n Indian Services companies have non-Indian leaders! With Biden-Harris ushering in continuity in visa and H1B policies favoring the human-talent spectrum, it is again proven that fairness, merit and creative pursuits in any field cannot be confined within any gender, caste, creed, race or nations' boundaries. Men (and women) are born free, and we don't have to necessarily bind ourselves in our own minds' chains as we grow up.

The Tech and Commercial worlds are future-proofing themselves. Like the world of Nature is doing it, even if in an apparently ruthless but inevitable manner. Well-governed, ESG-aware AI-automation policies both at national and international levels as well as at enterprise and business domains, will balance out the human-nature equations too. Let us nurture that hope and work towards realizing it for our next-gens.