Artificial Intelligence and the Future of Humans

Many of these experts have high hopes for continued incremental advances across all aspects of health care and life extension. They predict a rise in access to various tools, including digital agents that can perform rudimentary exams with no need to visit a clinic, a reduction in medical errors and better, faster recognition of risks and solutions. They also worry over the potential for a widening health care divide between those who can afford cutting-edge tools and treatments and those less privileged. They express concerns about the potential for data abuses such as the denial of insurance or coverage or benefits for select people or procedures.

Leonard Kleinrock, Internet Hall of Fame member and co-director of the first host-to-host online connection and professor of computer science at the University of California, Los Angeles, predicted, "As AI and machine learning improve, we will see highly customized interactions between humans and their health care needs. This mass customization will enable each human to have her medical history, DNA profile, drug allergies, genetic makeup, etc., always available to any caregiver/medical professional that they engage with, and this will be readily accessible to the individual as well. Their care will be tailored to their specific needs and the very latest advances will be able to be provided rapidly after the advances are established. The rapid provision of the best medical treatment will provide great benefits. In hospital settings, such customized information will dramatically reduce the occurrence of medical injuries and deaths due to medical errors. My hope and expectation is that intelligent agents will be able to assess the likely risks and the benefits that ensue from proposed treatments and procedures, far better than is done now by human evaluators, such humans, even experts, typically being poor decision makers in the face of uncertainty. But to bring this about, there will need to be carefully conducted tests and experimentation to assess the quality of the outcomes of AI-based decision making in this field. However, as with any 'optimized' system, one must continually be aware of the fragility of optimized systems when they are applied beyond the confines of their range of applicability".

Kenneth Grady, futurist, founding author of the Algorithmic Society blog and adjunct and advisor at the Michigan State University College of Law, responded, "In the next dozen years, AI will still be moving through a phase where it will augment what humans can do. It will help us sift through, organize and even evaluate the mountains of data we create each day. For example, doctors today still work with siloed data. Each patient's vital signs, medicines, dosage rates, test results and side effects remain trapped in isolated systems. Doctors must evaluate this data without the benefit of knowing how it compares to the thousands of other patients around the country (or world) with similar problems. They struggle to turn the data into effective treatments by reading research articles and mentally comparing them to each patient's data. As it evolves, AI will improve the process. Instead of episodic studies, doctors will have near-real-time access to information showing the effects of treatment regimes. Benefits and risks of drug interactions will be identified faster. Novel treatments will become evident more quickly.  Doctors will still manage the last mile, interpreting the analysis generated through AI. This human in the loop approach will remain critical during this phase. As powerful as AI will become, it still will not match humans on understanding how to integrate treatment with values. When will a family sacrifice effectiveness of treatment to prolong quality of life? When two life-threatening illnesses compete, which will the patient want treated first? This will be an important learning phase, as humans understand the limits of AI".

Charles Zheng, a researcher in machine learning and AI with the National Institute of Mental Health, commented, "In the year 2030, I expect AI will be more powerful than they currently are, but not yet at human level for most tasks. A patient checking into a hospital will be directed to the correct desk by a robot. The receptionist will be aided by software that listens to their conversation with the patient and automatically populates the information fields without needing the receptionist to type the information. Another program cross-references the database in the cloud to check for errors. The patient's medical images would first be automatically labeled by a computer program before being sent to a radiologist".

A professor of computer science expert in systems who works at a major U.S. technological university wrote, "By 2030 … physiological monitoring devices (e.g., lower heartbeats and decreasing blood sugar levels) could indicate lower levels of physical alertness. Smart apps could detect those decaying physical conditions (at an individual level) and suggest improvements to the user (e.g., taking a coffee break with a snack). Granted, there may be large-scale problems caused by AI and robots, e.g., massive unemployment, but the recent trends seem to indicate small improvements such as health monitor apps outlined above, would be more easily developed and deployed successfully".

Kenneth Cukier, author and senior editor at The Economist, commented, "AI will be making more decisions in life, and some people will be uneasy with that. But these are decisions that are more effectively done by machines, such as assessing insurance risk, the propensity to repay a loan or to survive a disease. A good example is health care: Algorithms, not doctors, will be diagnosing many diseases, even if human doctors are still 'in the loop'. The benefit is that healthcare can reach down to populations that are today underserved: the poor and rural worldwide".

Gabor Melli, senior director of engineering for AI and machine learning for Sony PlayStation, responded, "My hope is that by 2030 most of humanity will have ready access to health care and education through digital agents".

Kate Eddens, research scientist at the Indiana University Network Science Institute, responded, "There is an opportunity for AI to enhance human ability to gain critical information in decision-making, particularly in the world of health care. There are so many moving parts and components to understanding health care needs and deciding how to proceed in treatment and prevention. With AI, we can program algorithms to help refine those decision-making processes, but only when we train the AI tools on human thinking, a tremendous amount of real data and actual circumstances and experiences. There are some contexts in which human bias and emotion can be detrimental to decision-making. For example, breast cancer is over-diagnosed and over-treated. While mammography guidelines have changed to try to reflect this reality, strong human emotion powered by anecdotal experience leaves some practitioners unwilling to change their recommendations based on evidence and advocacy groups reluctant to change their stance based on public outcry. Perhaps there is an opportunity for AI to calculate a more specific risk for each individual person, allowing for a tailored experience amid the broader guidelines. If screening guidelines change to 'recommended based on individual risk,' it lessens the burden on both the care provider and the individual. People still have to make their own decisions, but they may be able to do so with more information and a greater understanding of their own risk and reward. This is such a low-tech and simple example of AI, but one in which AI can – importantly – supplement human decision-making without replacing it".

Angelique Hedberg, senior corporate strategy analyst at RTI International, said, "The greatest advancements and achievements will be in health – physical, mental and environmental. The improvements will have positive trickle-down impacts on education, work, gender equality and reduced inequality. AI will redefine our understanding of health care, optimizing existing processes while simultaneously redefining how we answer questions about what it means to be healthy, bringing care earlier in the cycle due to advances in diagnostics and assessment, i.e. in the future preventative care identifies and initiates treatment for illness before symptoms present. The advances will not be constrained to humans; they will include animals and the built environment. This will happen across the disease spectrum. Advanced 'omics' will empower better decisions. There will be a push and a pull by the market and individuals. This is a global story, with fragmented and discontinuous moves being played out over the next decade as we witness wildly different experiments in health across the globe. This future is full of hope for individuals and communities. My greatest hope is for disabled individuals and those currently living with disabilities. I'm excited for communities and interpersonal connections as the work in this future will allow for and increase the value of the human-to-human experiences. Progress is often only seen in retrospect; I hope the speed of exponential change allows everyone to enjoy the benefits of these collaborations".

An anonymous respondent wrote, "In health care, I hope AI will improve the diagnostics and reduce the number of errors. Doctors cannot recall all the possibilities; they have problems correlating all the symptoms and recognizing the patterns. I hope that in the future patients will be interviewed by computers, which will correlate the described symptoms with results of tests. I hope that with the further development of AI and cognitive computing there will be fewer errors in reports of medical imaging and diagnosis".

Eduardo Vendrell, a computer science professor at the Polytechnic University of Valencia in Spain, responded, "In the field of health, many solutions will appear that will allow us to anticipate current problems and discover other risk situations more efficiently. The use of personal gadgets and other domestic devices will allow interacting directly with professionals and institutions in any situation of danger or deterioration of our health".

Monica Murero, director of the E-Life International Institute and associate professor in sociology of new technology at the University of Naples Federico II in Italy, commented, "In health care, I foresee positive outcomes in terms of reducing human mistakes, that are currently still creating several failures. Also, I foresee an increased development of mobile (remote) 24/7 health care services and personalized medicine thanks to AI and human-machine collaboration applied to the field".

Uta Russmann, professor in the department of communication at FHWien der WKW University of Applied Sciences for Management & Communication, said, "Life expectancy is increasing (globally) and human-machine/AI collaboration will help older people to manage their life on their own by taking care of them, helping them in the household (taking down the garbage, cleaning up, etc.) as well as keeping them company – just like cats and dogs do, but it will be a much more 'advanced' interaction".

Lindsey Andersen, an activist at the intersection of human rights and technology for Freedom House and Internews, now doing graduate research at Princeton University, commented, "AI will augment human intelligence. In health care, for example, it will help doctors more accurately diagnose and treat disease and continually monitor high-risk patients through internet-connected medical devices. It will bring health care to places with a shortage of doctors, allowing health care workers to diagnose and treat disease anywhere in the world and to prevent disease outbreaks before they start".

An anonymous respondent said, "The most important place where AI will make a difference is in health care of the elderly. Personal assistants are already capable of many important tasks to help make sure older adults stay in their home. But adding to that emotion detection, more in-depth health monitoring and AI-based diagnostics will surely enhance the power of these tools".

Denis Parra, assistant professor of computer science in the school of engineering at the Pontifical Catholic University of Chile, commented, "I live in a developing country. Whilst there are potential negative aspects of AI (loss of jobs), for people with disabilities AI technology could improve their lives. I imagine people entering a government office or health facility where people with eye- or ear-related disabilities could effortlessly interact to state their necessities and resolve their information needs".

Timothy Leffel, research scientist, National Opinion Research Center (NORC) at the University of Chicago, said, "Formulaic transactions and interactions are particularly ripe for automation. This can be good in cases where human error can cause problems, e.g., for well-understood diagnostic medical testing".

Jean-Daniel Fekete, researcher in human-computer interaction at INRIA in France, said, "Humans and machines will integrate more, improving health through monitoring and easing via machine control. Personal data will then become even more revealing and intrusive and should be kept under personal control".

Joe Whittaker, a former professor of sciences and associate director of the NASA GESTAR program, now associate provost at Jackson State University, responded, "My hope is that AI/human-machine interface will become commonplace especially in the academic research and health care arena. I envision significant advances in brain-machine interface to facilitate mitigation of physical and mental challenges. Similar uses in robotics should also be used to assist the elderly".

James Gannon, global head of eCompliance for emerging technology, cloud and cybersecurity at Novartis, responded, "AI will increase the speed and availability to develop drugs and therapies for orphan indications. AI will assist in general lifestyle and health care management for the average person".

Jay Sanders, president and CEO of the Global Telemedicine Group, responded, "AI will bring collective expertise to the decision point, and in health care, bringing collective expertise to the bedside will save many lives now lost by individual medical errors".

Geoff Arnold, CTO for the Verizon Smart Communities organization, said, "One of the most important trends over the next 12 years is the aging population and the high costs of providing them with care and mobility. AI will provide better data-driven diagnoses of medical and cognitive issues and it will facilitate affordable AV-based paratransit for the less mobile. It will support, not replace, human care-givers".

John Lazzaro, retired professor of electrical engineering and computer science, University of California, Berkeley, commented, "When I visit my primary care physician today, she spends a fair amount time typing into an EMS application as she's talking to me. In this sense, the computer has already arrived in the clinic. An AI system that frees her from this clerical task – that can listen and watch and distill the doctor-patient interaction into actionable data – would be an improvement. A more-advanced AI system would be able to form a 'second opinion' based on this data as the appointment unfolds, discreetly advising the doctor via a wearable. The end goal is a reduction in the number of 'false starts' in-patient diagnosis. If you've read Lisa Sander's columns in the New York Times, where she traces the arc of difficult diagnoses, you understand the real clinical problem that this system addresses".

Steve Farnsworth, chief marketing officer at Demand Marketing, commented, "Machine learning and AI offer tools to turn that into actionable data. One project using machine learning and big data already was able to predict SIDS correctly 94% of the time. Imagine AI looking at diagnostics, tests and successful treatments of millions of medical cases. We would instantly have a deluge of new cures and know the most effective treatment options using only the data, medicines and therapies we have now. The jump in quality health care alone for humans is staggering. This is only one application for AI".

Daniel Siewiorek, a professor with the Human-Computer Interaction Institute at Carnegie Mellon University, predicted, "AI will enable systems to perform labor-intensive activities where there are labor shortages. For example, consider recovery from an injury. There is a shortage of physical therapists to monitor and correct exercises. AI would enable a virtual coach to monitor, correct and encourage a patient. Virtual coaches could take on the persona of a human companion or a pet, allowing the aging population to live independently".

Joly MacFie, president of the Internet Society, New York chapter, commented, "AI will have many benefits for people with disabilities and health issues. Much of the aging baby boomer generation will be in this category".
The overall hopes for the future of health care are tempered by concerns that there will continue to be inequities in access to the best care and worries that private health data may be used to limit people's options.

Craig Burdett, a respondent who provided no identifying details, wrote, "While most AI will probably be a positive benefit, the possible darker side of AI could lead to a loss of agency for some. For example, in a health care setting an increasing use of AI could allow wealthier patients access to significantly-more-advanced diagnosis agents. When coupled with a supportive care team, these patients could receive better treatment and a greater range of treatment options. Conversely, less-affluent patients may be relegated to automated diagnoses and treatment plants with little opportunity for interaction to explore alternative treatments. AI could, effectively, manage long-term health care costs by offering lesser treatment (and sub-optimal recovery rates) to individuals perceived to have a lower status. Consider two patients with diabetes. One patient, upon diagnosis, modifies their eating and exercise patterns (borne out by embedded diagnostic tools) and would benefit from more advanced treatment. The second patient fails to modify their behaviour resulting in substantial ongoing treatment that could be avoided by simple lifestyle choices. An AI could subjectively evaluate that the patient has little interest in their own health and withhold more expensive treatment options leading to a shorter lifespan and an overall cost saving".

Sumandra Majee, an architect at F5 Networks Inc., said, "AI, deep learning, etc., will become more a part of daily life in advanced countries. This will potentially widen the gap between technology-savvy people and economically well-to-do folks and the folks with limited access to technology. However, I am hopeful that in the field of healthcare, especially when it comes to diagnosis, AI will significantly augment the field, allowing doctors to do a far better job. Many of the routines aspects of checkups can be done via technology. There is no reason an expert human has to be involved in basic A/B testing to reach a conclusion. Machines can be implemented for those tasks and human doctors should only do the critical parts. I do see AI playing a negative role in education, where students may not often actually do the hard work of learning through experience. It might actually make the overall population dumber".

Timothy Graham, a postdoctoral research fellow in sociology and computer science at Australian National University, commented, "In health care, we see current systems already under heavy criticism (e.g., the My Health Record system in Australia, or the NHS Digital program), because they are nudging citizens into using the system through an 'opt-out' mechanism and there are concerns that those who do not opt out may be profiled, targeted and/or denied access to services based on their own data".

Valarie Bell, a computational social scientist at the University of North Texas, commented, "Let's say medical diagnosis is taken over by machines, computers and robotics – how will stressful prognoses be communicated? Will a hologram or a computer deliver 'the bad news' instead of a physician? Given the health care industry's inherent profit motives it would be easy for them to justify how much cheaper it would be to simply have devices diagnose, prescribe treatment and do patient care, without concern for the importance of human touch and interactions. Thus, we may devolve into a health care system where the rich actually get a human doctor while everyone else, or at least the poor and uninsured, get the robot".

The following one-liners from anonymous respondents also tie into the future of health care:

• "People could use a virtual doctor for information and first-level response; so much time could be saved!"
• "The merging of data science and AI could benefit strategic planning of the future research and development efforts that should be undertaken by humanity".
• "I see economic efficiencies and advances in preventive medicine and treatment of disease, however, I do think there will be plenty of adverse consequences".
• "Data can reduce errors – for instance, in clearly taking into account the side effects of a medicine or use of multiple medications".
• "Human-machine/AI collaboration will reduce barriers to proper medical treatment through better recordkeeping and preventative measures".
• "AI can take over many of the administrative tasks current doctors must do, allowing them more time with patients".