by Upasana Bhattacharjee
The International Conference for Automated Planning and Scheduling (ICAPS) hosts competitions every other year where computer systems try to find solutions to planning problems (such as scheduling flights). Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory are discovering processes to augment the technology by imbuing human intuition in them. This is another reminder of how far the developments in artificial intelligence have come along with their pace especially when posed against the economic progress and ethical understanding they demand.
Understanding the technology
Automated planning and scheduling is an aspect of artificial intelligence that is concerned with constructing strategies for machines (intelligent agents, autonomous robots, etc.) based on factors like the observability, determinism and variables involved in the situation. ICAPS is a forum for researchers and practitioners to ensure progress in the field. This branch of artificial intelligence is extremely useful in fields like space systems, software engineering and robotics.
Computers in the competitions organised by IPAC are given problems along with a range of difficulties following which the algorithm has to create a plan. The quality of plans is measured along a function that can gauge the level of efficiency of the solution.
36 MIT students were recruited to serve this end and given a set of planning problems. Their responses were then described in a formal language known as ‘linear temporal logic’.
In order to improve the algorithms that have proven to be less effective than high-functioning human planners and strategists in the past, insights from them and human intuition were encoded in machine-readable form and fed to the algorithms which resulted in a 10-15 per cent improvement in the algorithms’ problem-solving capacities. 36 MIT students were recruited to serve this end and given a set of planning problems. Their responses were then described in a formal language known as ‘linear temporal logic’. Efforts to convert inputs into linear temporal logic without human intervention are underway.
The contribution of scientists, intellectuals is pivotal for the evolution of society at large alongside robots Image courtesy: Pixabay
The way ahead
This development comes at a time when reports declare that 6 percent of the jobs in the US will be eliminated by robots by 2021. These developments demand the creation of new jobs and facilities to accommodate the displaced and work in this area is yet to catch up.
With AI now capable of recognising patterns in human intuition, human subjectivity could effectively be reduced to objective sets of binaries. As such, we must re-evaluate our understanding of what constitutes as human.
Artificial intelligence has seen enormous improvements and developments lately. The ability to recognise patterns and register human intuition is a big step in the field of AI and has significant ethical implications at play. We must make efforts towards understanding what the underlying ethical norm is with respect to AI. This ambitious field has attracted remarkable attention because of the intelligence it demands. The developments in the field have replicated human activity effectively along with reducing the possibility of mistakes. As such, they serve as rational replacements to human labour. We must then question if human intelligence should be left to prosper even when confronted by displacement in human lives.
While AI is a field that is seeing growth at a significantly high rate, it is a field that holds deep potential in terms of altering the society by demanding new economic policies along with a changed understanding of ethics and sociology. The developments in automated planning and scheduling by MIT are massive but the associated subjects to implement these developments with least disorder are still lagging behind. The change such technology promises will be mammoth, but if the social sciences do not catch up with them, the mammoth changes might result in more harm than gain.
No comments:
Post a Comment