Ever since we started relying on machines and automation, people have been worried about the future of work and, specifically, whether robots will take over their jobs. And it seems this worry is becoming increasingly justified, as an estimated 40% of jobs could be replaced by robots for automated tasks by 2035. There is even a website dedicated to workers worried about whether they could eventually be replaced by robots.

While machines and artificial intelligence are becoming more complex and, therefore, more able to replace humans for menial tasks, that doesnt necessarily apply to a wide number of industries. Here, well go through the sectors that continue to require the human touch.

Despite scientists best efforts, the language and translation industry cannot be replaced by machines. Currently, automatic translation programmes are being developed with deep learning, a form of artificial intelligence which allows the computer to identify and correct its own mistakes through prolonged use and understanding. However, this still isnt enough to guarantee a correct translation, as deep learning requires external factors, like language itself, to remain the same over time. As we know, language is constantly developing, often with changes so subtle, you cant tell its happening. For a machine to be able to accurately translate texts or speech, it would need to be constantly updated with every new modification, across all languages.

Machines are also less able to pick up on the nuances found in speech or text. Things like sarcasm, jokes, or pop culture references are not easily translated, as the new audience may not understand them. Translating idioms is a particularly common example of this, as these phrases are generally unique to their dialect. In the UK, for example, the phrase its raining cats and dogs means its raining heavily. You would not want this translated on a literal level. As London Translations state in an article on the importance of using professionals for financial text translation, literal translations are technically correct, but read awkwardly and can be difficult to comprehend due to poor knowledge of the source language. Needless to say, these issues would be totally unacceptable in a document as important as a financial report.

Translating with accuracy not only requires fluency in both languages, but also a complete understanding of cultural differences and how they can be compared. Machines are simply not able to naturally make these connections without having the information already inputted by a person.

Finding the perfect candidate for a role can get stressful, especially if you have a pool of excellent potential employees to choose from. However, there are now algorithms that recruiters can use to help speed the process up and, theoretically, pick the most suitable person for the job. The technology is being praised for its ability to remove discrimination, as it simply examines raw data, and thus omits any sense of natural prejudice. It can also work to speed up the hiring process, as a computer can quickly sift through applicants and present the most relevant ones, saving someone the job of having to manually read through every application before making a decision.

However, in practice, its not that simple. Recruiting the right candidate should be based on more than qualifications and experience. Personality, attitude, and cultural fit should also be considered when recruiters are finding a candidate, none of which can be picked up on by machines.

One way of minimising this risk could be to introduce the algorithm at an earlier stage, through targeted ads or to help sift through initial applications. This allows recruiters to look at relevant candidates, rather than those that wouldnt have passed the initial screening anyway. However, this could conversely work to introduce bias to the recruitment process. The Harvard Business Review found that the algorithm effectively shapes the pool of candidates, giving a selection of applications that are all similar, fitting the mould that the computer is looking for. The study found that targeted ads on social media for a cashier role were shown to 85% of women, while cab driver ads were shown to an audience that was around 75% black. This happened as the algorithm reproduced bias from the real world, without human intervention. Having people physically checking the applications can serve to prevent this bias, introducing a more conscious effort to carefully screen each candidate on their own merits.

More people than ever before are meeting their partners online, according to a study published by Stanford University. And while a matchmaking algorithm sounds like a dream for singletons, it doesnt mean that they are able to effectively set you up with your life partner. As these algorithms are actually the intellectual property of each app, Dr Samantha Joel, assistant professor at London, Canadas Western University, created her own app with colleagues. Volunteers were asked to complete a questionnaire about themselves and ideal partners, much like typical dating websites would. After answering over 100 questions, the data was analysed and volunteers were set up on four-minute-long speed dates with potential candidates. Joel then asked the volunteers about their feelings towards any of their dates.

These results then identified the three things needed to predict romantic interest: actor desire (how much people liked their dates), partner desire (how much people were liked by dates), and attractiveness. The researchers were able to subtract attractiveness from the scores of romantic interest, giving a measure of compatibility. However, while the algorithm could accurately predict actor and partner desire, it failed on compatibility. Instead, it may be worth sticking to the second most common way of meeting a partner through a mutual friend. Your friends will be able to make educated decisions about relationships, as they have a deeper understanding of preferences and compatibility in a way that a machine simply cant replicate.

Author Bio: Syna Smith is a chief editor of Business usa today. She has also good experience in digital marketing.

More:
The industries that can't rely on machine learning - The Urban Twist