The Future of Artificial Intelligence: To Kill or To Heal?by@mobidev
1,232 reads
1,232 reads

The Future of Artificial Intelligence: To Kill or To Heal?

by MobiDevApril 8th, 2020
Read on Terminal Reader
Read this story w/o Javascript
tldt arrow

Too Long; Didn't Read

The Future of Artificial Intelligence: To Kill or To Heal? The future of AI holds a great deal of promise for humanity and can change the world for the better. The use cases described in this article, not only give the understanding of how AI is currently being used, but also help to imagine its future potential. Facial recognition technology can play a vital role in public safety and public safety. Countries like China rely on this technology to maintain social control. This kind of use fuels the use of the technology.
featured image - The Future of Artificial Intelligence: To Kill or To Heal?
MobiDev HackerNoon profile picture

The evil cyber-intelligence from the Matrix and a cyborg killing machine from the Terminator movies - that’s what most people used to imagine when talking about the future of artificial intelligence.

Today’s artificial intelligence is incapable of taking over human jobs. It will not cause a robot-powered apocalypse. Ethicists and computer scientists don’t see doomsday scenarios in the future, either. The future of AI holds a great deal of promise for humanity and can change the world for the better.

The use cases, described in this article, not only give the understanding of how AI is currently being used, but also help to imagine its future potential.

Check the article to get a full understanding of where AI is heading these days: 10 AI and Machine Learning Trends To Impact Business in 2020

Social Cost of Biased AI

Current AI technologies are wrestling with errors caused by incomplete data sets. An area of major concern is bias in decision-making. These errors are being recognized and can be overcome.

Consider the US criminal justice system's history of biased human decision-making. The same bias exists in the financial industry and educational institutions. Machine learning models have repeated human biases.

The COMPAS system is an example of biased AI. Several states in the US used the software to determine the risk level of criminal offenders. COMPAS used a dataset with a small number of risk factors and features. The resulting algorithm disproportionately flagged Black defendants as being high risk. These false flags occurred at double the rate seen in white offenders. As a result, a disproportionate number of Black individuals were sentenced more often and more severely than whites.

Biased decisions also occur when AI models use limited visual data sets. For example, a face recognition system was trained only on Caucasian face datasets. It was applied to diverse populations in the real world.

Inaccuracies were rampant. African-American and Asian individuals were misidentified at rates of 10 to 100 times more often than Caucasians. The National Institute of Standards and Technology (NIST) found that many current face recognition systems are not usable in the fields of law enforcement and national security due to these discrepancies.

Data science engineers must consider training their models on high-variation datasets for face recognition app development to avoid bias.

Face Recognition Under the Ban Hammer

Face recognition seems to be a promising technology for government and enterprises, but no. For some reasons, society does not think so.

Oakland and San Francisco's police departments and public agencies ban the use of facial recognition tech. Police departments in Oregon, California and New Hampshire cannot use the technology in their police body cameras. One in four US residents believes that the federal government should restrict the use of facial recognition technology. Americans aren’t alone.

The EU’s General Data Protection Regulation (GDPR) has found that facial recognition does not meet consent requirements. In 2019, a Swedish school used the software in a limited way to track student attendance. Parents gave consent for biometric collection. Yet, the EU fined the school. DPA ruled that biometrics such as facial recognition are sensitive personal data. Under the regulation, sensitive personal data cannot be collected and stored, unless the person being undergone the recognition activity gives the consent.

Early in 2020, the European Commission debated a ban on facial recognition in public spaces. The proposed ban would last for five years. This ban never occurred. It was removed from the final iteration of the Commission’s White Paper on Artificial Intelligence.. 

Currently, US has its own local law, which is close to GDPR. The California Consumer Privacy Act mirrors the GDPR and is the default US standard for data privacy. Regulatory band and guidelines limit technological innovations in facial recognition.

Innovators are developing facial recognition software despite the limitations set by consumer groups and government agencies. The tech is gaining popularity in China and the US. In 2019, the facial recognition market was valued at $5.07 billion (US). By 2025, economists expect a $10.19 billion (US) value, calculated at a CAGR of 12.5% over a five-year span (2020-2025). Forecasters expect facial recognition technology to enter the retail market where it will predict sales and personalize shopping experiences.

Another Side of the Technology

Modern surveillance infrastructure requires facial recognition technology. Countries like China rely on this technology to maintain social control. This kind of use fuels the cry for bans and facial recognition restriction. Yet, facial recognition is a multi-use technology that can serve other purposes. Facial recognition technology can play a vital role in public health and safety.

Governments and private industries are using facial recognition technology to suppress and contain COVID-19. Agencies can require area residents or employees to provide their travel history, their name, identification number and their temperature before scanning a QR code. Once scanned, access is granted to an area. Also, social media platforms are tracking movements and have enabled hotlines for users to report the illness. This information provides information on hotspots and regional spread.

In China, facial recognition technology detects elevated temperatures in a crowd of people. It also can detect whether a person is wearing a mask. Reports show that facial recognition technology can identify mask-wearing people with a 95% accuracy rate. Consumer-facing apps capture personal health information to provide users with information about the health of those close by.

The Chinese public was introduced to facial recognition technology after the 2008 Olympics in Beijing. It came into wide use across government agencies, financial services and retail sectors after 2015. There has been some concern about its widespread use. The public appears to accept its practical use and in some instances appreciate its ability to contain COVID-19 transmission.

Revealing the Danger

Natural language processing use cases are another example of how artificial intelligence will change the future. Prior to the rapid spread of COVID-19 into countries outside of China, NLP technology provided insight that possibly saved lives.

BlueDot is an AI platform that uses NLP and machine learning to track infectious diseases across the globe. It does this by employing algorithms that rapidly browse a multitude of sources. The algorithms are designed to flag early signs of epidemics. In the last weeks of December 2019, the platform recognized a cluster of “unusual pneumonia” diagnoses in Wuhan, China. A little over a week later, the World Health Organization (WHO) came out with an official statement on the existence of a “novel coronavirus” in a patient in Wuhan.

BlueDot isn’t the only AI that can flag areas of concern across thousands of sources. Alibaba, a global E-commerce powerhouse, created StructBERT, which is powered by NLP models. The models are capable of processing viral gene-sequences at a fast rate, as well as screening proteins. Alibaba has put the platform to use in the fight against COVID-19. It is freely available to researchers and scientists who can use the information and technology to speed the development of vaccines.

Finding the remedy

An example of how artificial intelligence will change the future of public health and biotech is seen in two separate pre-print publications from February 2020. 

One paper conveys how South Korea-based Deargen developed and implemented a machine-learning model. The model identified four possible antiviral medications that could mitigate the effects of COVID-19. Deargen's MT-DTI is a learning model that relies on chemical sequences and not on 2D or 3D molecular structures. The model predicts if a molecule of interest, like a virus, will bind to a target protein. Atazanavir is an anti-HIV drug that has been approved by the FDA. MT-DTI found that atazanavir could bind to a protein located on the outside of a SARS-CoV-2 molecule and block its ability to bind to human proteins. SARS-CoV-2 causes COVID-19.

Hong Kong-based Insilico Medicine also published a pre-print paper in February. Insilico used an AI-based platform to model thousands of novel molecules. The system searched for a novel molecule that could disrupt SARS-CoV-2 replication.

These firms are only two of a multitude of tech firms, academic research labs and government scientists working towards a remedy. An open dataset was created to help communication between groups. The COVID-19 Open Research Dataset, or CORD-19, is a central hub for all research on the subject. Its data is machine-readable and constantly updated. Anyone accessing the database can take the stored data and easily apply it to machine learning models and AI technology. This can only speed the rate of research currently being performed across the globe.

Future of AI in the post-COVID-19 World

The future of artificial intelligence is dependent on human management and creative solutions. Scientists and researchers hold the big picture - the cure - in their minds. Human creativity takes machine-parsed data and puts it together in meaningful ways. Without the right human input, AI is an efficient data collector and not much more. The COVID-19 crisis highlights the truth about AI. Innovators determine AI's usefulness and value when they set the models in motion.

What Decision Makers Can Do Now

The pandemic will end, sooner or later. The economic impact of the virus is yet to be determined. But there are several scenarios for the economic impact of the COVID-19. 

Decision-makers can choose to evolve, innovate and embrace new technology. All you have to do is take a fresh look at the situation. That’s when we  can help to guide your vision into successful business outcomes.

Click to know how to evolve your software product through the crisis.