Myths about Artificial Intelligence (AI)
Information & Communications Technology and Media | 24 Oct 2019
“I think the danger of AI is much greater than the danger of nuclear warheads by a lot and nobody would suggest that we allow anyone to build nuclear warheads if they want,” said Elon Musk.
Myth #1: AI will take over humans and our jobs
AI will replace human jobs
AI has the potential to seriously disrupt labour and already has in many areas. Many fear AI will replace them in the workplace but forget that AI is meant to work with humans. This is more than a straightforward transfer of labour from humans to machines. Technologies like the barcode scanner and ATM machines sparked fears of unemployment when introduced. Since then, they have improved their respective industries and created new types of jobs.
Employers generally look at AI as a method of augmenting human workforces and enabling them to work in new and smarter ways. AI will change some current job categories, create entirely new roles, and allow employees to work in a more efficient and clever manner.
AI will only replace low-skilled and manual jobs
AI already carry out some work reserved for highly trained professionals such as doctors and lawyers. Much focus has been on the day-to-day aspects of the work, e.g. in law, AI is used to scan thousands of documents at lightning speed, drawing out points which may be relevant in an ongoing case. In medicine, machine learning algorithms have shown high competence in assessing images from scans and x-rays for warning signs of disease. In the financial and insurance industry, roboadvisors are used for wealth management and fraud detection.
AI enables businesses to make more accurate decisions based on predictions, classifications and clustering, enabling AI solutions to reach deep into work environments, replacing mundane tasks and augmenting the more complex ones. Many professions involve a combination of routine and “human touch” procedures. AI does not eliminate the human involvement in these tasks but may eventually limit it to observing and dealing with unusual cases. This means the need to adjust job profiles, capacity planning and offer retraining options for existing staff. For instance, a lawyer will hone the way he argues and presents a case, and for doctors it could be breaking news in the most compassionate and helpful way.
Myth #2: AI has human characteristics, including thinking and learning like a human brain
AI is a computer engineering discipline consisting of software tools aimed at solving problems. Some forms of AI might give the impression of being clever, charming or feature aspects of emotion or personality, like iOS Siri and Amazon Alexa. However, these responses are not organic and are programmed so. This is a misconception brought about when we think of intelligence as linear and maybe because of how pop culture portrays AI.
In reality, intelligence is measured in many different dimensions, where e.g. in speed of calculations or capacity for recall, computers far outpace us, while in others such as creative ability, emotional intelligence and strategic thinking, they are nowhere near.
Although AI today solves one task exceedingly well, it fails if the conditions of the task change, e.g., while image recognition technology is more accurate than most humans, it cannot solve a math problem. A finished machine learning product gives the impression that it is able to learn on its own, but we should remember that to enable the AI to function, experienced human data scientists need to frame the problem, prepare the data, determine appropriate datasets, remove potential bias in the training data and continually update the software to enable the integration of new knowledge and data into the next learning cycle.
Myth #3: AI is totally objective
AI technology is based on data, rules and other kinds of input from human. As all humans are intrinsically biased, so will the AI. Systems that are frequently retrained, e.g. using new data from social media, even more vulnerable to unwanted bias or intentional malicious influences.
An increasingly common and problematic misunderstanding is that the only prerequisite to AI success is lots of data. Linden says. “In addition to technological solutions such as diverse datasets, it is crucial to also ensure diversity in the teams working with the AI and have team members review each other’s work. This simple process can significantly reduce selection and confirmation bias.”
There are jobs in AI and machine learning teams now that are focused almost entirely on curating and cleaning data. “It’s not the quantity of training data that matters, it’s the quality,” says Rick McFarland, chief data office at LexisNexis Legal and Professional. “Large volumes of poorly or inconsistently labelled training data don’t bring you closer to an accurate outcome. They can trick modelers by creating “precise” results since the formula for variance is inversely dependent on sample size. In a nutshell, you get precisely inaccurate results.” (link)
Myth #4: My business does not need an AI strategy
Every organisation should consider the potential impact of AI on your strategy and how this can be applied to your business. In many ways, avoiding AI is the same as forgoing the next phase of automation and could place enterprises at a competitive disadvantage. “Even if your current AI strategy is ‘no AI’, this should be a conscious decision based on research and consideration. As with every other strategy, it should be periodically revisited and changed according to the organization’s needs,” says Linden.
AI is already present in many aspects of our lives (link to previous article), so much so that we do not even notice it. In the same vein, implementation of AI doesn’t always require substantial expert research and investment. What it boils down to is your business needs and strategy.
Final Thoughts: AI has been around for a long time and it is not going anywhere. And that is a good thing as AI has so many positive benefits for the human race.