6 Types of Disadvantages Of AI

Disadvantages Of AI

As artificial intelligence plays an increasingly more significant role in everyday life, its potential effect on society and our daily lives is an issue we all need to know and discuss.

What are the disadvantages of artificial intelligence?

Artificial intelligence has a profound effect on society, an impact that promises to be even more significant as the technology becomes more sophisticated. But not everything is guaranteed to be positive.

We’ve put together a list of our 7 disadvantages of artificial intelligence that we should all be aware of.

1. Unemployment

With growing fears that automation and artificial intelligence will change the way we work and force people into unemployment, questions remain raised about which machines will replace jobs in the future. Some experts point to potential changes in occupations by 2030, estimating that 75 million to 375 million workers (3 to 14 percent of the global workforce) will need to change jobs and learn new skills.

It shows a wide prediction gap, ranging from optimistic to very pessimistic. It highlights that many experts in the technology and business sectors do not share a standard view of the future of our labor market. In short: it’s hard to say how many jobs will disappear.

2. Lack of transparency

Al can remain flawed in many ways, which is why transparency is critical. The input data may be full of errors or poorly cleaned. Or perhaps the data scientists and engineers training the model inadvertently selected biased datasets in the first place. In typical application development, quality assurance and testing processes and tools can quickly detect potential bugs. But with so many things that could go wrong, the real problem is a lack of visibility: not knowing why the AI ​​is malfunctioning. Or sometimes not even that it performs poorly.

But AI isn’t just code, and the underlying models can’t just remain examined to see where the bugs are – some machine learning algorithms are inexplicable, kept secret (because it’s in the business interests of their makers), or both. It leads us to a limited understanding of the biases or errors that AI can cause. Courts have begun implementing algorithms in the United States to determine a defendant’s “risk” to re-offend and inform bail, sentencing, and parole decisions. The problem is that there is little oversight and transparency about how these tools work.

3. Biased and discriminatory algorithms

It brings us to our next topic. “Unbiasedness” is not only a social or cultural problem. It also occurs in the technical sphere. Design errors or inaccurate and unbalanced data fed into algorithms can lead to biased software and engineering artifacts. Thus, AI merely reproduces the racial, gender, and age biases that already exist in society and exacerbates social and economic inequalities. You probably read about Amazon’s hiring experiment a few years ago.

The tool used artificial intelligence to find candidates by ranking them from one to five stars – similar to how shoppers rate products on Amazon. It was discriminatory against women because Amazon’s computer models remained trained to screen applicants by following patterns in resumes submitted to the company over ten years, effectively favoring male candidates and penalizing resumes that contained the word “women.”

4. Profiling

AI can remain used to create frighteningly accurate profiles of people. Algorithms remain developed to find patterns, so when they tested their ability to collect personal data in a competition, they remained shown to be able to predict a user’s likely future location based on past location history. The prediction was even more accurate when using location data from friends and social contacts. Sometimes this disadvantage of artificial intelligence is downplayed. You might think you don’t care who knows your moves. After all, you have nothing to hide.

First, it’s likely not entirely true. Even if you don’t do anything wrong or illegal, you may not want your personal information to be available in its entirety. After all, you wouldn’t move into a house with transparent walls. So is it really that you don’t care about sharing your device’s location history? What about your teenage daughter’s location history? Would you be comfortable with someone releasing her location data, including predictions? Certainly not. Information is power, and the information we give up is power over us.

5. Disinformation

The rise of misinformation is a downside of artificial intelligence that we are already witnessing. In 2020, the Activist group Extinction Rebellion created a deep fake to create a fictional speech by Belgian Prime Minister Sophie Wilmès.

Unfortunately, this was not the only case. Deepfakes will gradually be used for targeted disinformation campaigns, threatening our democratic processes and causing social polarization.

Adding to these misinformation problems are online bots that can generate fake text, including news articles altered to encourage views of hoaxes or tweets. An AI language tool, GPT-3, recently produced tweets saying, “They can’t talk about warming because it’s not happening anymore,” to generate skepticism about climate change.

6. Environmental impact

Although AI can positively impact the environment, for example, by enabling intelligent grids to match electricity demand or enabling innovative and low-carbon cities. One of the disadvantages of artificial intelligence is that it can also cause significant damage to the environment due to its intensive use of energy. A 2019 study found that a specific type of AI (deep learning in natural language processing) has a huge carbon footprint due to the fuel the hardware requires.

Experts say that training a single AI model produces 300,000 kg of CO2 emissions, roughly equivalent to 125 round-trip flights from NYC to Beijing, or five times the lifetime emissions of the average (US) car. And, of course, model training is not the only source of emissions.

The carbon footprint of the infrastructure surrounding big-tech AI implementation is also significant: data centers must remain built, and the materials used need to be mined and transported.

Conclusion

Companies that grab AI startups globally are dangerous because, as a result, they will play too much of a role in determining the direction AI technology takes. With dominance in search, social media, online retail, and app stores, these companies have near monopolies on user data. They are becoming the primary suppliers of AI to everyone else in the industry. Such a concentration of power is dangerous because it risks giant tech companies dictating democratically elected governments.

Also read: Types of AI and what are they?