Brave New World

As in past years, the UNESCO Netexplo Innovation Forum 2019 in Paris identified a number of interesting and innovative ideas from startups and research labs all over the world. They have a network of academics (yours truly is one of them) whose students spot some 2,000 innovative (digital or socially valuable) ideas that have bubbled up in the last year or two, and after considerable analysis, the top 10 are invited to present their ideas at the event. The audience, which consists mostly of managers from European firms, then votes one of them to be the Grand Prix winner.

My students focus on innovations from Asia, especially from India. As in past years, one of their choices reached the final 10, but, alas, did not win the grand prize. The company that was identified by us was the intriguing Israeli firm D-ID, that does de-identification of photographs so that, while easily recognised by humans, the digital images are difficult for automated systems to process, as their pixels have been modified. In an era of privacy concerns about ubiquitous surveillance systems, this is a useful tool.

The general trends identified from the spotted innovations (Netexplo is at pains to emphasise these are merely current trends, and not a forecast of which firms will do well in future) by technology expert Sandrine Cathelet also fall into related areas of automation.

In past years, the trends identified included panopticon-like systems that subjugate the populace through systematic surveillance. In passing, I visited the Cellular Jail (Kaala Pani) in Port Blair earlier this year, and it was eerie that the principles of the panopticon — a device proposed by Jeremy Bentham, a supposedly liberal British philosopher, to break the will of those the state wishes to terrorise — were fully implemented in the design of that structure.

Incidentally, San Francisco passed a law in early May that bans the use of facial recognition by government agencies: the backlash against the panopticon is beginning. There are fears about both Big Brother (as seen in China) and about racial discrimination (algorithmic and data-set bias means darker-skinned people are much more prone to being mis-identified as criminals).

In addition to the innovation grand prize, this year Netexplo branched off into a new area: smart cities. Although we have become accustomed to that phrase being thrown around casually as marketing jargon, the vision outlined at Netexplo is quite comprehensive: indeed, it is a startlingly broad coming-together of scientific, technological, ethical, environmental and human factors concerns that we will surely hear more about going forward: the smart city as, in a way, a metaphor for civilisational progress (with all its attendant unintended consequences).

Going back to the innovations, the trends identified this year also offer some cause for concern. For, they are about the potential implications of artificial intelligence (AI). Today’s robots are beginning to be (monstrously) efficient. Systems are beginning to be chimeras these days: there are bionic eyes and bio-hybrid robots, examples of hyper-symbiosis, with machines and ‘cyborgs’ sharing data.

Clearly, there are positive stories: for instance, AI software can now predict cancer in tumors based on image recognition better than human radiologists can. A recent trial in breast cancer by MIT and a nearby hospital demonstrated this. But there are questions: one is the data set used to train (inadequate representation across races, geographies, etc) and the second is the nagging fear that all this is based only on correlation — of histories of people who developed cancers post-screening. And we know that correlation is not the same as causation.

Monstrously efficient cyborgs, hyper-present, hyper-sensitive, and hyper-efficient, can control the environment quickly and effectively. These could be chameleons that can face any situation at all, drawing on resources from the digital world. As an example, finding a mate need no longer be left to chance: Pheramor, a dating application, tries to figure out your compatibility with a potential mate based on physical chemistry and social alignment. Matchmakers, you are out of luck in this Orwellian world!

Besides, the more friendly and human-like the robots get — and we had examples of these in the top 10 winners’ ideas — the more we will be seduced into accepting them as, in fact, human. There is the interesting story of Eliza, a 1980s-era (and quite primitive by today’s standards) AI system that acted like a psychologist, rephrasing things that you typed in and turning them into questions, slowly drawing you out. It turned out that people would pour their hearts out to this system, even more so than to a human psychologist, perhaps because it appeared non-judgmental. But that system was merely emitting canned responses. What if today’s equivalent AI truly understands you?

That could be the next phase, of monsters from AI: ghosts in the machine. We could slowly be turned into machines, at least in part. We could be downgraded and excluded in society if we have lost certain skills and capabilities that we have outsourced to the cyborg. The question could well become: what is the value of a human? Indeed, there is a large question about what happens to humans who become superfluous, as they are no longer needed? Will we enjoy a life of leisure, or will we be ‘sacrificed’ by monstrous AIs? The ethics of AI could be an existential threat for the entire human race.

There are many unresolved questions here, and we need to consider not only how to future-ready ourselves, but also how to robot-proof the future. Big issues, with no simple answers, and good questions to cogitate on: cogito, ergo sum.

1. Furhat: a Swedish humanoid robot face with social skills, which appears to listen, has more-or-less human facial expressions, and even blinks like humans. This could enable the use of more friendly robots.

2. Factmata: as India’s elections have demonstrated, fake news is an ever-present danger. This British application, a browser add-on, creates a fact-checking community, and reframes the information presented to identify sexism, racism, propaganda, fake news, and to warn the user.

3. TCAV: one of the serious concerns about deep learning is that it’s a ‘black box’: the algorithm does not tell us how it arrived at the ‘weights’ it imposed on various parameters while ‘learning’, and then used for its predictions. This US project from Google Brain is a ‘translator for humans’ and could well be an important part of the AI revolution, as it demystifies AI decision making, thus alleviating concerns about bias.

4. I-cut: five girls aged 15 to 17 from Kenya developed this app for those in danger of female genital mutilation so they can reach out to the authorities, and medical and legal help for those who have already been mutilated. The Grand Prix winner!

5. Soundshirt: a German app that turns music into vibrations you feel as each instrument will be felt on the skin using the product. You ‘hear’ the music in different points on your body. Though developed for deaf people, it is enjoyable to even people who can hear.

6. Spatial. This American application creates immersive shared workspace with Hololens virtual reality headsets. The idea is to provide the next level of immersion compared to Telepresence, and thus reduce the time, expense and global warming related to business travel, without losing the immediacy of engagement.

7. AI Biodiversity monitor. A British Raspberry Pi-based hardware device, powered by solar energy, is intended to continuously track biodiversity in tropical countries. Using its microphone and camera, it can record data about the natural evolution of the rainforest, as well as provide early warning of problems.

8. D-ID. An Israeli application protects your privacy from increasingly intrusive facial recognition systems by modifying photographs so that machines cannot decipher them.

9. rAInbow. A South African chatbot that helps victims of domestic violence to identify the type of abusive relationships they are in and to seek help before it is too late. It can help victims get away from a Stockholm Syndrome-like blaming of themselves for toxic relationships, and connect them to volunteers and authorities.

10. Biohybrid robot. A Japanese advancement that could be a major stepping stone to a full-fledged cyborg. Using muscles from rat stem cells grown in situ, the researchers have built a small cyborg over a robotic skeleton and that can simulate a grasping hand. These are living muscle cells applied to biohybrid sensors, reactors, actuators and processors. The image I walked away with was that of ‘Blade Runner’. We may not be too far away from that vision!

The cyborgs may indeed be coming, and also maligning artificial intelligences. So this could be a brave new world or one where humans are in severe, existential conflict with cyborgs: science fiction in general has predicted the latter, going back all the way to Metropolis to 2001: A Space Odyssey. We shall see which vision prevails but it has implications within our lifetimes.

To be parochial for a moment, it was a little disappointing that no Indian innovation made it to the top 10, and very few into the top 100. Hearteningly, though, Indian-origin people in other countries were responsible for three or four of the top 10. There are also statistics that the Regeneron Science Talent Search in the US, one of the premier contests of ingenuity, has almost 40 per cent of its participants being Indian-American. Maybe Indian science and innovation are doing all right, except perhaps not in India.