Jan 10, 2018 | Stuart Frankel
2018 Predictions on Artificial Intelligence
Artificial Intelligence infiltrated the headlines in 2017, dominating conversations that spanned a wide variety of topics such as election results, chatbots, and ethical frameworks. Though the hype of AI will begin to subside in 2018, its adoption will continue to grow.
Thus, we focused our 2018 predictions on the ways in which AI will continue to become a normal part of everyday living, from regulations to frustrations. Take a look at our predictions for the year to come.
Prediction 1: The term “Artificial Intelligence” will become obsolete as AI becomes invisible.
As our interactions with AI-powered devices, virtual agents, and smart machines continue to grow, we will stop asking “Is this AI?” and focus instead on the value received from those interactions.
The new question will be: Was my customer service issue resolved quickly, and has the chatbot appropriately and efficiently answered my questions? Not: Was that chatbot an AI? From Nest devices that can send smoke alarms to RPAs that can synthesize disparate data to fuel automated reporting systems, intelligent systems are becoming increasingly interconnected and working behind-the-scenes to make previously mundane tasks smarter.
Just as we stopped using the term “Big Data” when we realized it’s just data, we’ll stop using “Artificial Intelligence” when we realize it just works intelligently.
Prediction 2: Regulations around AI will be instituted, requiring transparency into AI systems’ decision-making.
Ethical discussions surrounding AI have continued to heat up, with a steady influx of headlines focused on how technology is interfering with election results, disrupting the stock market, and pushing inherent biases. Tech giants have formed cross-company initiatives to ideate ethical frameworks, as well as instituted internal boards to closely examine their own product development efforts.
The next step will take shape in the form of government regulation, which may result in laws that stipulate AI systems be able to explain all of the decisions they have made in a transparent and straightforward manner. "My AI did it" should not excuse illegal behavior.
Prediction 3: VC investment in AI will hit an inflection point.
Venture capitalist investment in AI start-ups has continued to hit record highs, and they aren’t the only ones investing. Per a recent McKinsey report, tech giants like Google spent nearly $30B on AI in 2016, with 90% devoted on R&D and 10% on AI acquisitions. An “AI” bubble is clearly upon us, but will it burst akin to the dot-com boom of the nineties?
We say, not yet. The hype behind broad AI applications will subside as companies start to demand returns on their AI investments and expand their adoption of AI technologies, particularly into vertically-oriented solutions, to advance their business.
Prediction 4: Expectations and frustrations in AI will hit an all-time high.
The magic of AI will begin to wear off as everyone starts to realize that AI is not magic. In fact, it can take organizations massive amounts of resources in time, capital, and talent to get an AI project into production, often with unclear ROI (remember the unsuccessful $62 million MD Anderson investment in IBM Watson?).
Instead of hiring Chief AI Officers and formulating broad innovation initiatives, companies will learn the three critical factors of ensuring operational success with their AI investments: establish use cases that will deliver a clear return on investment, ensure the technology can integrate seamlessly into existing workflows, and require that your AI delivers transparency in its decision-making.
Prediction 5: The most-in-demand job of 2018 will be the Data Engineer.
A few years ago, the Data Scientist was chosen as the Sexiest Job of the 21st Century. Fast-forward to now, when data scientists are, of course, still valuable, but the dependency on them as the sole interpreter of key analytical insights has waned.
With AI, we now have scalable solutions that can analyze, interpret, and communicate data insights across a company, allowing data scientists to spend time on higher value tasks. This is giving rise to a new ‘hot’ job in the enterprise: the data engineer. This individual sources, curates, and integrates available data so it can be easily leveraged by various tools and personnel.
A Year In Review: How did our 2017 Predictions Fare?
Now, let’s look back on last year’s predictions to see what actually came true in 2017 and what has not yet been fully realized. Let’s review.
2017 Prediction 1 - The movement towards conversational interfaces will accelerate.
Nailed it! Use of conversational interfaces has greatly increased over the past year, fueled by the rise of chatbots and the popularity of Amazon’s Alexa (over 11 million Echoes were sold last holiday season).
However, enterprise adoption of conversational interfaces has proven to be more challenging than consumer adoption, as these intelligent agents are currently more skilled at task-oriented activities (ie: “Order pizza, Alexa!”) than business-oriented ones (ie: “Alexa, why are my sales declining?”).
These interfaces will not only need to chat with users, but will also need to learn and understand the context of business domains in order to become more widely used outside the home.
2017 Prediction 2 - Design will begin evolving to increase our trust in AI.
Heading that direction- There is still a lot of work to be done to increase people’s trust in AI, but the conversation has definitely been top-of-mind, with increased discussions regarding how to implement an ethical framework and how to design systems that elicit trust and offer transparency. More research is still needed to determine how to regulate AI and what the universal standards of AI should be - tech giants will continue to lead the conversation.
2017 Prediction 3 - We’ll start talking about how AI systems talk to each other.
Nailed it! There has certainly been more discussion on how AI systems communicate and the possibilities of machine-to-machine interactions. From the proliferation of self-driving cars to tightly integrated smart home devices that can detect everything from motion to temperature changes, even water leaks, AI-enabled IoT technologies are operating autonomously through their ability to connect with each other and gain a better sense of the world around them.
2017 Prediction 4 - Discussions of imbedded bias in AI will grow.
Nailed it! In 2017, there were multiple cases of bias within AI systems, as the systems are only as good as the data used to teach to them. The first step to combating bias is acknowledging it as an issue. Initiatives like Google’s PAIR aim to take a step in the right direction, but it is likely that combating bias will be an ongoing effort within the AI community.
2017 Prediction 5 - Enterprises will start to demand ROI from their AI.
Heading that direction- Despite an onslaught of funding for AI companies, it’s difficult to see which ones will grow into successful businesses. Referring to a product as “powered by AI” may help with fundraising, but that doesn’t mean it is delivering ROI to those who purchased it. Enterprises are taking note of this. In order to adequately track ROI, organizations should choose problems they can solve with demonstrable results and ensure their AI solutions can be integrated into existing workflows.