Once we used to marvel at robots and other contraptions in science fiction movies, or in comic strips that stretched our imaginations. Few ever thought about any of these appearing in real human lives, at least not in our lifetimes. Fast forward not so many years, and lo behold, they, or their construction, seem to be ubiquitously sprouting, even seemingly sneering at those they will displace. From such innocuous gadgets like the pocket calculator, the automated teller machines (ATMs) taking our commands and dispensing the desired money-amount, and trapping the human voice and functions for automated reproduction, we now find some of these have evolved into far more sophisticated beings, outperforming humans in speed, precision, quality, and output, even commanding us what to do.
Representing the sputtered start of the Fourth Industrial Revolution that the World Economic Forum (WEF) addressed in its 2017 Davos gathering, artificial intelligence (AI) does not represent one equipment but many, addressing the output multiplication problem industries face, facilitating the recognition and knowledge functions humans utilise to claim progress, and the (human) substitution possibility that ripples so fearfully across assembly-lines and the production manager's office. At once the very AI advent both emboldens and encloses us: how we use what we use to do what we want depends, in part, on the infrastructural sophistication of the community we live in (do we have the assembly-lines to robotise, or a large enough market to even think of output multiplication, or even the university where the requisite training can be done); and, in part, on how intellectually skilled and/or portable we are (to either lock ourselves up in an ivory-tower producing necessary software, or move to where our skills may be most in demand).
With its abundant population and lower technology demand, Bangladesh exemplifies countries where any AI evolution will be long drawn-out and initiated more through copy-cat than frontier piercing exercises, like expanding production. Without any workforce infrastructure to uphold worker's rights, for example, adopting AI contraptions remains the Damoclean Sword for disruptive workers and their unions: Bangladesh's largest exporting industry, the ready-made garment, can easily be automated should work-related costs or obstacles increase, just as spiralling wages elsewhere brought that industry into Bangladesh in the mid-1970s.
On the other hand, an infrastructural and industrially sophisticated country like the United States seems littered with AI accessible windows: from expanding production to even substituting teachers in class, a wide-range of functions and jobs can be undertaken and completed mechanically with far superior results than expected from human faculties. In between lies a long list of countries with AI exposure, some experimenting at the low-end, others pushing the high-end farther. How soon they plunge into any of these AI formats becomes a spin-off of technological capabilities (as a launching pad, improvisation tool, or frontier-scraping function), economic resources (measuring efficiency), social adaptability (whether a labour trade-off, or a culture-emasculating demon), and political preferences (job-creation policies, and so forth).
Once calibrated, the next step would be to determine what type of an AI contraption is needed. Plenty of work is being done in this regard. Thomas Davenport and Rajeev Ronanki, for example, identified "three business needs" for which AI gadgets have evolved (Harvard Business Review, January-February 2018, 108-16), which they call (a) process automation; (b) cognitive insight; and (c) cognitive engagement. One can deduce how the requisite technologies constitute a staircase of sophistication, with the typical starting point being robots, drones, or some such other mechanism, to expand production and boost precision, in other words, serve as a speedier replacement of physical functions. Then the shift is made to reproduce mental functions, with cognitive skills performing lesser or greater roles: the former learns from human functions, becoming 'analytics on steroids', and dealing with data-intensive tasks, much as the pocket calculator had started to do over half-a-century ago; beyond this 'cognitive insight' (or machine learning or deep learning), lies the fertile, breathtaking, and fear-triggering 'cognitive engagement' AI typology, the arena where substituting the human mind threatens human replacement most egregiously, for example, automated bank-tellers or surgeons.
From interviewing 250 corporation executives, Davenport and Ronanki discovered how the top priority of businesses has thus far been less to reduce human workers (only 22 per cent supported this), than to expand output, but to do so incrementally, not dramatically (51 per cent). If process automation gets priority, cognitive counterparts breed other problems: forbidding costs (40 per cent), inadequate understanding of the know-how (37 per cent), inadequate available talent (35 per cent), and a technological disinclination (18 per cent).
Of course, few of these observations would be relevant to countries like Bangladesh at present; but that they expose kernels of growth, breakthroughs, literary enhancements, and cataclysmic outbursts would probably not keep such information in the background for too long. If anything we know about AI possibilities, it is that any surge in AI usage can be spontaneously sudden, therefore imperative for us to know about them before we get swamped or disrupted by them. Addressing an entirely different issue in the same Harvard Business Review, Robert S. Kaplan, George Serafeim, and Eduardo Tugendhat convey a terribly important message. "Instead of trying to fix local problems," they argue, "corporations and other actors need to reimagine [sic] the regional ecosystem in which they participate if they are to bring poor farmers and unemployed urban youths into the mainstream economy".
That might be an agenda we can handle, indeed must, if we remain true to our middle-income identity. Instead of boosting production or displacing restless workers, a combination of the 'process automation', 'cognitive insight', and 'cognitive engagement' AI gadgets can help us address and redress our growth-constraining elements: poverty, food self-sufficiency, renewable fuel supplies, corruption, traffic congestion, and so forth.
In other words, our technological sophistication will be most successful the faster and more completely we forge the right social contract. Ultimately that becomes a political function: stability, policy continuity, and resource allocation expose some of the critical building-blocks. The moment we are confident we have positioned them, the faster we can unfasten our seat-belt and plunge into the technological wilderness.
Dr. Imtiaz A. Hussain is Professor & Head of the newly-built Department of Global Studies & Governance at Independent University, Bangladesh.