Why AI Still Needs a Human Touch

People have been talking about an AI takeover since at least 1872 (gee, thanks, Samuel Butler). We’ve seen multiple examples of “the future” in media that convinced us we’d all have robot assistants by now. But the stark reality is, even in 2021, AI still needs a human touch.

What Is AI?

As a company that offers a Conversational AI platform, we’re used to people asking us, “How would you even use AI in… texting?” And that’s a fair question. Because AI is considered fashionable, it’s become an overused buzzword — extremely popular in marketing. And because of this, the actual term “AI” has lost some of its clarity.

Just look at how many definitions there are:

  • The ability of a computer or computer-controlled robot to perform tasks commonly associated with intelligent beings (Britannica).
  • The simulation of human intelligence in machines that are programmed to think like humans and mimic their actions (Investopedia).
  • A field, which combines computer science and robust datasets, to enable problem-solving (IBM).
  • The science and engineering of making intelligent machines (Stanford).

Though these definitions have some similarities, some specifically mention “robots,” while others refer to a field or area of technology. It’s no wonder why some of us are confused about AI.

Here’s all you need to know — AI is anything that can imitate intelligence, with the operative word being imitate.

How AI Learns (and Why There Won’t Be an AI Uprising Anytime Soon)

An AI is only as good as the data it learns from (that’s why some investors make the mistake of funding companies and data scientists who don’t have a significant body of data). But if you’re from outside the tech field, you might not know how an AI learns from data.

AI learns from repetition. The more times it observes something occurring, it learns and becomes able to imitate it. If this process sounds relatable, DON’T WORRY! You’re not a robot. The way an AI learns is very similar to how humans learn as children — much of it comes from repeating observed behavior.

Take Tesla’s self-driving cars, for example. The data used for the AI to self-drive has been collected over years. In the beginning, it was limited to recording lane assist, and then little by little, it was able to handle more. Soon it was able to recognize stop signs, then display a stop sign, and even stop at the sign. In the future, it might even be able to handle taking off after the car is stopped at a stop sign.

AI is nice to have because it makes our lives more convenient, but it’s not going to take over the world anytime soon. For now, it still needs a human touch to handle it. After all, the human brain is the most complex thing ever researched. That’s why we’re not discounting the value of people when AI is involved.

AI Needs Humans… For Now

AI needs humans because it’s not perfect yet. For an AI to develop, it needs repetition and training provided by humans. And if something happens that confuses it, an AI will also need a human to take over for extra assistance and to provide quality assurance. There is, however, a point where the AI surpasses the need for human assistance — when it achieves narrow intelligence in a subject.

No one thought an AI could beat human chess master Garry Kasparov, but once the AI got enough data and reached narrow intelligence, it was able to beat him. (According to Wired, Garry doesn’t hold a grudge, though.) AI has also triumphed over human players in poker and even more complex games like StarCraft II and Pokémon. So, if an AI can master games that require a fair amount of intelligence, will Drips’ AI eventually be able to master conversational messaging without human assistance? Absolutely.

Even though we’ve spent years getting data from humans to make our AI more advanced, the way humans communicate is highly nuanced and occasionally needs a human to translate a meme or sequence of emojis. Now, out of half a million conversations, Over 99% of the touchpoints are from our AI in a completely humanized fashion. We wouldn’t have been able to do this without data and repetition with a human touch.

TL;DR, if something can be automated, it should be. But if you can leverage AI with human QA, then you’re still in a good place.

Want to see Drips’ AI-powered (human backed!) platform in action?

Schedule a Demo

Load More
Share This Post :