Although AI has been lauded as revolutionary and game-changing, it is not without flaws.As AI becomes more powerful and prevalent, the voices warning about the potential risks of AI become louder.
"The emergence of artificial intelligence may herald the end of the human race," Stephen Hawking said.
This is not an isolated concept for the renowned theoretical physicist.
Unease abounds on several fronts, including the rising automation of some jobs, gender and racially-biased algorithms, and autonomous weapons that function without human oversight (to mention a few). And we're still in the early phases of discovering what AI is truly capable of.
ARTIFICIAL INTELLIGENCE RISKS
- Job losses as a result of automation
- Invasion of privacy
- Deepfakes
- Bad data causes algorithmic bias.
- Inequality in society
- Market turbulence
- Automatization of weapons
Concerns regarding who is developing AI and for what goals make it even more important to comprehend its possible drawbacks. We will look at the potential threats of artificial intelligence and how to mitigate them in the sections that follow.
1) JOB LOSSES AS A RESULT OF ARTIFICIAL INTELLIGENCE AUTOMATION
As AI-powered job automation becomes more prevalent in industries such as marketing, manufacturing, and healthcare, it is becoming a significant concern. Between 2020 and 2025, 85 million jobs are predicted to be lost to automation, with Black and Latino workers being particularly vulnerable.
According to futurist Martin Ford,
"the reason we have a low unemployment rate, which doesn't really
represent those who aren't seeking work, is partial because this economy
has spawned a lot of lower-paying service sector jobs." "I don't
believe that will go on,"
As AI robots improve in intelligence and dexterity, the same tasks will require fewer people. While it is true that AI will create 97 million new jobs by 2025, many individuals will lack the technical skills required for these roles and may be left behind if corporations do not upskill their workforces.
If you're now flipping burgers at
McDonald's and more automation comes in, will one of these prospective careers
suit you? Apparently, says Ford. Or is it feasible that the new position may
need substantial education, training, or even inborn abilities that you lack,
such as great interpersonal skills or ingenuity? Because those are the things
that computers are not particularly good at, at least not yet."
Even occupations requiring graduate degrees and other post-college training are vulnerable to AI displacement.
Law and accounting, for example, are ready for an AI takeover, according to technology consultant Chris Messina. In fact, according to Messina, some of them may be decimated. AI has already had a substantial impact on medicine. Law and accounting are up next, according to Messina, with the former undergoing a "major shakeup."
"Think about the complexities of contracts, and truly diving in and knowing what it takes to build the perfect transaction structure," he added of the legal industry. "Many attorneys are combing through a lot of information - hundreds or thousands of pages of data and records. It's quite easy to overlook something.
So AI that can comb through and completely give the best possible contract for the outcome you're looking for is likely to replace a lot of corporate attorneys."
2) SOCIAL MANIPULATION USING ARTIFICIAL INTELLIGENCE ALGORITHMS
One of the main hazards of artificial intelligence, according to a 2018 assessment of potential abuses, is social manipulation. This concern has become a reality as politicians rely on platforms to promote their views, with Ferdinand Marcos, Jr., most recently using a TikTok troll army to win the votes of younger Filipinos in the 2022 election.
TikTok is powered by an artificial
intelligence algorithm that floods a user's feed with information linked to
prior media they've viewed on the platform. The app's critics focus on this
procedure and the algorithm's failure to filter out harmful and inaccurate
content, casting doubt on TikTok's ability to protect its users from dangerous
and deceptive media.
Deepfakes invading political and social arenas have made online media and news even murkier. The technology makes it simple to swap out the image of one figure in a photograph or video. As a result, bad actors now have another channel for spreading misinformation and war propaganda, creating a nightmare scenario in which distinguishing between credible and flawed news might be practically impossible.
No one is aware of what is genuine and what isn't, according to Ford. "Therefore, it truly results in a scenario in which you can't rely on what we've historically thought to be the best possible evidence, you can't even accept what your own eyes and ears are telling you. It is going to be a huge issue."
3) SOCIAL SURVEILLANCE IMPLEMENTED BY AI TECHNOLOGY
In addition to the more existential issue, Ford is concerned about how AI will affect privacy and security. A classic example is China’s usage of facial recognition technology in companies, schools, and other locations. In addition to tracking a person's whereabouts, the Chinese government may be able to collect enough data to track a person's activities, relationships, and political opinions.
Another example is police forces in the United States using predictive policing algorithms to predict where crimes will occur. The issue is that these algorithms are driven by arrest rates, which disproportionately affect African American areas. Police agencies then increase their presence in these communities, raising concerns about over-policing and whether self-proclaimed democracies can avoid using AI as an authoritarian tool.
4) THE INCREASE IN SOCIOECONOMIC INEQUALITY AS A RESULT OF AI
Companies that refuse to admit the inherent biases built into AI algorithms risk jeopardizing their DEI projects via AI-powered recruiting. The notion that AI can assess a candidate's characteristics through facial and voice analysis is nonetheless polluted by racial biases, duplicating the same discriminatory hiring practices that firms claim to be eliminating.
Another source of concern is the widening socioeconomic inequality caused by AI-driven job loss, which reveals the class biases of how AI is implemented. Automation has resulted in salary decreases of up to 70% for blue-collar workers who do more physical, repetitive activities. Meanwhile, white-collar workers have mostly been unaffected, with some even earning higher earnings.
Claims that AI has somehow broken down social barriers or provided more jobs do not provide a clear picture of its consequences. It is critical to take into account disparities based on race, class, and other factors. Otherwise, determining how AI and automation benefit some people and groups while harming others becomes more challenging.
5) ARTIFICIAL INTELLIGENCE-BASED BIAS
AI bias in many forms is also harmful. Professor Olga Russakovsky of computer science at Princeton claims that prejudice in AI goes well beyond issues of race and gender.AI is built by humans — and humans are inherently biassed — in addition to data and algorithmic bias (the latter of which can "amplify" the former).
"A.I. researchers are mostly men, from specific racial demographics, who grew up in high socioeconomic areas, and primarily people without disabilities," Russakovsky explained. "Since we're a rather homogeneous population, it's difficult to think generally about global issues."
The lack of expertise among AI
inventors may explain why speech-recognition AI frequently fails to recognize particular dialects and accents, or why businesses fail to consider the
repercussions of a chatbot impersonating infamous characters in human history.
Developers and corporations should take greater care to prevent reinforcing
entrenched biases and prejudices that endanger minority groups.