Chatbots don’t have the best reputation.

From creating their own secret languages to spouting racist and sexist comments to simply not being very helpful, the emerging technology still has its very public shortcomings.

While increasingly being used by businesses for customer service, the technology is still in its rough, early stages, Forrester vice president Julie Ask said. But 2018 could be the year when the technology finally begins to mature.

“(Chatbots) certainly have very limited functionality,” Ask said, adding that they’re a “work in progress.” “Most are hard-coded decision trees. Most need to escalate to humans very quickly to do anything more than answering a simple question,” she said.

SEE: IT leader’s guide to the future of artificial intelligence (Tech Pro Research)

The bots are only beginning to be leveraged for business purposes, and are currently used in information distribution and guided decision-making situations, as well as for marketing, Ask said.

Fiverr, which began a chatbot subcategory in July 2017, found that 55% of customers looking for help with chatbots were doing so to better engage customers via social media. Assistance with customer service, online shopping, and collecting lead information were other top needs.

Overcoming barriers for developers

Chatbots hold the promise for natural language processing (NLP), or being able to converse with human customers in a naturalistic way, said Warner Goertz, research director at Gartner.”That’s the target trajectory we’re on, but we’re not there yet,” Goertz said.

To hit that target, the technology must overcome a series of barriers. One is developing the platform itself, as using a third party may lead to data privacy issues, Ask said.

Simple parts of the chatbot interaction need to be clarified, Forrester analyst Andrew Hogan said. For example, companies should select things the bot can do well, and make it clear what it cannot do, Hogan said. The bot should also know when to end a conversation or when to continue helping the person it’s talking to, Hogan said.

User experience needs to come first when it comes to chatbot creation, with the technology coming second, said Rob Harles, managing director at Accenture Interactive. Developers should work to make the chatbot seem as human and conversational as possible, with consistent updates.

“A chatbot is like a baby–it needs to be nurtured and taught–you can’t just set up a chatbot and let it go,” Harles said.

Developers should avoid getting too attached to a certain technology, and think about the reasoning for a chatbot, Harles said. The bot should set out to solve a problem for the customer, making it a useful, uncomplex tool.

“Some organizations think chatbots will automate a process or avoid call center calls–but the truth is chatbots are like the advent of the ATM,” Harles said. “Many thought the ATM would replace banks and tellers, but in reality, it simply created a new channel. That is what chatbots will deliver for brands–a new communication channel.”

But a new communication channel can come with its own issues, including potential abuse of the channel with inappropriate or unrelated questions. Jordi Torras, Inbenta CEO, noticed four themes of strange chatbot questions developers and brands should be prepared to deal with: Personal questions, current affairs, not-safe-for-work content, and seemingly random questions.

A bot that appears human–with a name and face–tends to feel more personal for users, Torras said, which can be better for business and the user experience. However, human-appearing bots seem to receive more NSFW remarks. Torras recommends developers set up a way to defuse or ignore the situation, and to offer chat guidelines at the beginning of all conversations to deal with unrelated questions.

Business appeal

Once bots overcome these obstacles, or at least make significant progress, their business appeal and usage may increase. While customer service and social media guidance leading chatbot adoption now, there are still other ways businesses can use bots.

Some companies use the technology internally, to increase efficiency and free up employee time. uses a bot to answer internal sales questions, increasing efficiency and automating over 56% of responses. Sales increased by 14.4% after implementing the bot, according to CEO Cristian Rennella. And Freelock Computing‘s bot helps the company by running timers, sending reminders, and providing login links.

Others use the Facebook chatbot–which over two-thirds of business professionals on Fiverr said is the ideal chatbot platform–as part of their company instead of only for customer service. For example, HelloAva, a skincare app, provides product and skincare regimen recommendations after a customer takes a quiz through a Facebook chatbot.

A few companies use chatbots as the basis of their business. One company,, replaces websites with conversational bots. Another, Heek, uses a bot to guide SMBs through creating their own website.

Will bots replace human jobs?

In some cases, human jobs may be able to be replaced by a chatbot. Customer service representatives are one role that may be replaced or refocused, as are web agencies or site designers.

“Previously, when you wanted to make a website with an agency, you would have to create a brief, explain the brief to an account manager who will then brief the designer and the developer,” Heek CEO Nicolas Fayon said, adding that chatbots can decrease time and cost, while freeing up these people for larger projects.

Additionally, if tech jobs like site designers can be automated, people with those skills may be free to fill other tech roles that are facing a tech skills shortage, like software engineers.

But the technology will need human help to get through its growing pains and into a breakout year.

“You need to have humans behind these technologies to better ‘adjust’ how chatbots interact with humans, and correct bias,” Fayon said.

To start using chatbots in 2018, Goertz recommends businesses begin trials and proof of concept studies now.

“Amazon introduced Alexa For Business in November 2017–this is an effective, inexpensive and resource-friendly way to get systems started,” Goertz said. “Machine learning capabilities will then take this starting base and automatically improve it over time, using artificial intelligence (AI) capabilities.”