It’s incredible how far AI, in particular, Conversational AI, has come in the last decade. Ten years ago, the names Alexa and Siri might have referred to children in your child’s class; today, they play an important role in organizing our lives. From finding directions to our next meeting to ordering dog food, our reliance on AI-powered virtual assistants is only increasing as the technology improves day by day.

While we’re still in the early days, there’s no denying the promise of AI in the future. So, as we enter a new decade, Finn AI’s executive team thought it would be fun to predict what the future will hold for conversational AI.

From efficiencies in how we build and deploy AI to how we both enable and manage it, here are a few of our top 10 conversational AI predictions for 2020:

  1. We’ll gain efficiencies in AI optimization
  2. There’ll be a shift in focus to hybrid models
  3. Explainable AI will be key to building trust
  4. A new strategic role will emerge: the AI Translator
  5. 2020 will separate the wheat from the chaff in AI
  6. There’ll be more international coordination on AI principles
  7. New initiatives will support the commercialization of AI for good
  8. We’ll chat with support…but not on the phone
  9. The margin crunch in retail banking will see more turn to AI
  10. We’ll get closer to a Universal Language Translator

 

AI TECHNOLOGY TRENDS

By Dr. Kenneth Conroy, VP of Data Science at Finn AI
@cionad / @kenneth-conroy

Conversational AI Changing

 

1. We’ll Gain AI Processing Efficiencies

2019 was the year of the global climate strike movement. In 2020, sustainability and carbon reduction will be top of mind in every industry — AI is no exception.

Minute gains in AI performance come with exponential processing costs and use an incredible amount of resources. A report by Cornell University found that common AI training models (Transformer, ELMo, BERT, and GPT-2) are so energy-intensive, they can emit more than 626,000 pounds of carbon dioxide equivalent. That’s about five times the lifetime emissions of the average car, including the manufacture of the car itself! The models’ power usage increased drastically when extra training was added to increase accuracy.

As AI developers hone AI model performance in 2020 and beyond, we expect them to focus on creating efficiencies and finding innovative ways to improve performance and accuracy without the increased use of resources.

2. There’ll Be a Shift in Focus to Hybrid Models

Rather than pure deep learning models, we expect to see a shift toward hybrid models in the coming year. According to Forbes, the hybrid approach to development will be even more productive as a hybrid natural language model combines the advantages of linguistic and machine learning.

We also expect to see more focus on domain expertise when designing solutions for specific problems. For example, we’ll see more industry-specific or use case-specific chatbots that can go deep on certain subjects, hold advanced conversations, and take more complex actions than those we’ve seen to date.

 

AI BEST PRACTICES TRENDS

By Steven Zhu, VP of Technology, Finn AI
@stevencanhelp

AI best practices

 

3. Explainable AI Will Be Key To Building Trust

Explainable AI (XAI) refers to AI practices whereby the results of the AI can be understood and justified by human experts. It is the antithesis to black box AI where even the designers cannot explain why the AI arrived at a specific decision.

According to AI experts at Georgian Partners, XAI will help increase consumer trust in AI. However, others argue that transparency rarely comes free and that there will be trade-offs between the accuracy and the explainability of a solution. That being said, we expect explainability to play a leading role in AI best practices in 2020.

 

COMMERCIAL TRENDS

By Stephen Menon, VP of Product, Finn AI
@stephenmenon001

Commercial Trends

 

4. A New Strategic Role Will Emerge: The AI Translator

We predict that 2020 will see the rise of the AI translator in many organizations. This senior executive-level role sits at the intersection of business strategy and AI methods and, according to Harvard Business Review, is responsible for translating strategic objectives and business models into the types of AI that can advance them. 

This role will not be easy to fill as it requires a business leader who has an understanding of business models, code development, algorithm creation, and product development and who can bridge the gap between the business needs and the technology. This person will also be responsible for explainable AI practices.

5. 2020 Will Separate the Wheat from the Chaff in AI

Gartner predicted that by 2020, the average person would have more conversations with bots than with their spouse. This didn’t come to pass (for most of us, hopefully!), since chatbots didn’t live up to the hype. However, industries like fintech, healthcare, and retail are adopting the technology with great success. 

For conversational AI companies, 2020 will be a make-or-break year. There’s no more room for false starts. Customers demand excellence and won’t put up with poor chatbot experiences from brands they trust. If an AI product does not create real, repeatable value for users, it’s unlikely that it will exist in 2021.

GLOBAL REGULATORY TRENDS

By Natalie Cartwright, Co-Founder & Chief Operating Officer, Finn AI
@nataliecartwright

Global AI Trends

 

6. There’ll Be More International Coordination on AI Principles

In 2020, we expect to see stronger international coordination and collaboration about the principles and legal frameworks governing the ethical use of AI.

Examples of this include the recently announced Canadian and French-led Global Partnership for AI (GPAI), the increased United Nations AI leadership, and the OECD Principles on Artificial Intelligence that promote AI that is innovative and trustworthy and that respects human rights and democratic values.

7. New Initiatives Will Support the Commercialization of AI for Good

We predict that 2020 will bring the launch of new policies, tools, and initiatives that support the democratization and commercialization of AI for good. 

We will see more initiatives like the data trust proposed by the Open Data Institute. A data trust is a legal structure that provides independent stewardship of data for the benefit of all. 

We will also see more knowledge networks like AI Commons that will allow anyone, anywhere to benefit from the opportunities that AI can provide. These networks identify and connect communities around problem-solving with AI and offer access to shared resources, frameworks, and collaboration opportunities.

LEADERSHIP / LONG-TERM TRENDS

By Jake Tyler, Co-Founder & CEO, Finn AI
@jaketyler

Happy with the use of AI

 

8. We’ll Chat With Support…But Not on the Phone

Let’s face it, phoning customer support is a terrible experience for all but the most tricky problems. In 2020, customers don’t want to run through an IVR or sit on hold for all eternity. 

Customers want to chat…by text, not by phone.

We expect that up to 20% of all our interactions with brands will happen via messaging in the coming year, whether it’s chat on websites, in apps or in third-party channels like iMessage and WhatsApp.

9. The Margin Crunch in Retail Banking Will See More Turn to AI

In 2020, we’ll continue to see margin pressure in retail banking as challenger banks start to gain traction in the US and big tech enters the market. 

We saw broker margins drop to zero in 2019 and we expect this to extend elsewhere as new entrants creep in and target the most profitable business lines first. These entrants run a materially lower-cost model, making more meaningful money elsewhere. This won’t play out in full in 2020 but we’ll start to see the writing on the wall.

10. We’ll Get Closer to a Universal Language Translator

We end our predictions with a big, bold, exciting one. It will likely come to pass later in the decade but it’s one to watch.

It relates to one of the very first goals of artificial intelligence research: machine translation (translating one human language into another). While we’ve made strides in this area, we’re nowhere near the goal of simultaneous real-time translation whereby, as a speaker is talking in one language, the AI automatically outputs a translation in another language so that the two speakers can carry on a natural dialog.

According to Marc Andreessen at A16z, we will soon get to a stage where we have a universal language translator. This will create a huge reduction in barriers to global business and unlock a huge talent pool of highly-skilled people who didn’t speak English.

 

That wraps up our predictions for 2020. We’d love to hear your thoughts. Connect with us on social media to comment on our AI predictions or share some predictions of your own.

Finn AI
Founded in 2014, Finn AI is the world’s leading AI-powered conversational banking technology provider, working with top financial institutions including: ATB Financial, Banpro Grupo Promerica, Fidor Bank, and TymeBank, as well as partnerships with MX, Auth0, Liverperson, Temenos, and Visa. Banks and Credit Unions use the award-winning Finn AI platform to transform and deepen customer engagement--providing a truly personalized digital-first experience--while delivering the operational efficiencies and cost savings of conversational AI.