Does AI need a soul? | NTT DATA

Thu, 31 January 2019

Does AI need a soul?

An important analogy coined by Andrew NG (the artificial intelligence guru) states that "Artificial Intelligence is the new electricity". Evidently, the effect of AI on all activity sectors in society is comparable (if not greater) than the effect of electricity on the 'second industrial revolution'.

As shown in the following illustration published in the US media in 1889, the perception of technology in that specific time period can be equated to a certain "technophobia" or "technoscepticism", which is recently being observed around AI.

 

Several leading experts in the field of AI agree that the technology is still in the very nascent stages of development and is far from the point of achieving universal artificial intelligence. However, this constitutes a key moment in which a society must establish a clear framework with respect to ethical responsibility and the social impact of artificial intelligence. Thus, it is important to explore the following question- Why are ethics so relevant in the development of artificial intelligence?

The starting point corresponds to the fact that any technology is neutral until we (as humans) determine its use. However, artificial intelligence exhibits characteristics that make it especially dangerous if it is used for unethical purposes: 

  • Scale: when artificial intelligence is combined with other powerful technologies (such as social networks), the resultant algorithms can significantly affect large social groups. Examples of this include recent scandals that enveloped Facebook and Cambridge Analytica with respect to the alleged manipulation of electoral results.
  • Secretism: Currently, several algorithms on which artificial intelligence development is based rely on neural networks. A disadvantage of neural networks is that they behave in a manner similar to black boxes. We know the inputs and outputs that an algorithm receives and generates, however it is not possible to explain the reasoning followed by AI to reach the recommendations that are proposed. This is especially sensitive when a certain algorithm (voluntarily or involuntarily) causes a bias with ethical or social derivatives in its results. In the aforementioned cases, it is extremely difficult to establish automatic detection and feedback mechanisms to alleviate the problem. An extreme example of what constitutes a black box is Google's artificial intelligence that can detect the gender of an individual from just a single image of the retina with a very high success rate. To date, doctors perform this classification in a completely random manner since humans (at this point) are unaware of the AI pattern(s) to determine whether an individual is male or female based on a photograph of their retina. Given the aforementioned points, artificial intelligence presents a high risk of negatively impacting society if it is used on a massive scale and if attention does not focus on the data with which it is trained. A representative (albeit fortunately controlled) example occurred when Microsoft published its virtual assistant Tay.ai on twitter. In less than a day, and by "learning" from the comments of the people who interacted with it, Tay acquired the racist biases of certain individuals in an extremely short time period given the scale of worldwide interactions.

Global response is necessary for a society to decrease the impact of the lack of ethics in AI

In a previous post, I analyzed different public and private initiatives to create ethical frameworks with respect to artificial intelligence development and indicated that several challenges posed by artificial intelligence on society can only be resolved via global collaboration.

In summary, this is a key point in time for artificial intelligence development with real scientific results and an immense potential in direct business applications. Nevertheless, it is simultaneously accompanied by certain immaturity, excessive expectations, and knock-on effects of social impact and ethics of the technology that are paramount. Will artificial intelligence transform companies and society? Undoubtedly, the answer is a resounding yes. Will it happen ethically and contribute towards the development and improvement of companies and society? Certainly, the human brain is in the driver’s seat on these matters (at least for now).

 


How can we help you

Get in touch