By Felipe Cornejo
Artificial intelligence (AI) may seem like the newest and hottest trend coming out of Silicon Valley, but this type of technology has actually been around for quite a while. The first work done in this area dates back to the 1950s. The early work done on AI focused on translation. Throughout the cold war, IBM created systems to translate Russian to English. Ever since then, there have been advances in this fields along with boost and boom cycles, known as AI Winters. We can find a few examples of AI technologies in the 1970s and 1980s, such as micro-world and expert systems. Today, AI is more accessible to all kinds of organizations. Understanding its evolution allows me and my team to create better services and solutions to our clients and partners.
I think we are currently undergoing another hype cycle. However, this time might be different due to three factors: advancements and breakthroughs in machine learning, data available to train models, and computational power. Such progress has motivated corporations and startups to make significant investments in this technology. Hence, AI solutions are becoming more ubiquitous in our daily operations. Small-business leaders should understand the implications, how to implement this technology, and the types of use cases artificial intelligence can help streamline -- along with the benefits it can bring to their organizations.
So, what is machine learning and how does it impact general AI? Machine learning is an AI technique that aims to get computers to behave like humans. Within this technique, deep learning, an algorithm where data structures are modeled on the human brain, is making a lot of progress.
Data is the new oil. Its quantity and quality continue to improve. Thanks to this massive amount of datasets available, deep learning models are getting more sophisticated and reliable. The last piece of the puzzle is computing power. Thanks to Moore’s law (the notion that computational power doubles every 18 months), we have now access to more powerful and cheaper processing capacity to train deep learning models and implement artificial intelligence solutions.
Like hardware, the cost of AI development is coming down as the supply of professionals and development tools increase. We have seen this phenomenon before, with the commoditization of web development. The access to skills and tools represents a huge opportunity for organizations looking to innovate. Recent research from McKinsey found that 45 percent of work activities could potentially be automated by today’s technologies and 80 percent of that is enabled by machine learning. For quick wins, focus on internal processes and pain points. For example, focus in areas where there are a lot of interaction between computers, APIs or systems. Also, think about areas where algorithms can have a direct impact on revenue, production and cost. Once quick wins and support from the organization are consolidated, start looking at the front of the house and customer-facing touch points and interactions.
In the past year, my company has been helping some of our clients implement custom AI solutions. Through this process, we have also seen use-cases that can be applied more broadly in different industries. Here are a few examples.
- Customer Service: Currently, one of the more traditional use cases for AI is customer support and service. Companies have been improving their customer support processes and customer experience through the use of chatbots. The idea of using virtual customer representatives is twofold: decrease cost on the back of the house while providing a more tailored and personalized assistance to your constituents.
- Hiring: Hiring is another area where automating processes and utilizing existing AI solutions can help leaders streamline their processes and make better decisions. Case in point, applications leverage natural language processing techniques in order to improve the quality of new hires. These solutions can help analyze the applicant responses and profile information in order to assess whether he or she will be a culture fit for the organization. At my organization, we use these tools to make better assessments on the candidate's emotional quotient and other trades that we think are important to fit in our culture.
- Training: Organizational knowledge and training are areas that one might not associate right away with artificial intelligence, but this is another instance where AI can improve work. Organizations can implement systems that assist with the training and coaching of the workforce, and help improve forecasting. For example, Udacity was able to improve sales by 50 percent when it introduced chatbots to its sales team. These chatbots were able to coach the salesperson and provide information on what sets of words, conversations and information led to more success in closing sales.
The last few years we have seen tremendous progress in algorithms, data and infrastructure for artificial intelligence, and the cost of acquiring them is decreasing. Today these tools are not only available for the enterprise but also for smaller organizations. It's important for small and midsize businesses to understand the impact of this technology on their organizations and industries. At its core, AI is deep automation of processes, tasks and work, so small business should not be afraid of implementing these types of projects.
Felipe Cornejo is co-founder and CEO of Devsu.