AI is disrupting businesses process through small operational algorithms. AI for businesses has been in focus in 2019, from reducing human exposure to dangerous elements to eliminating repetitive tasks.
According to Webroot, 71% of the US enterprises are planning to leverage more AI/ML tools for security. Still, currently, only 49% of the IT professionals are feeling extremely comfortable using such devices. The change might be overwhelming for many of the current security experts; that’s why about 76% of the employees don’t care if their business uses them.
According to a recent survey by Adobe, currently, 90% of the IT leaders are forecasting the use of AI/ML increase in the future, while 41% of them are looking for technology that is currently powered by AI. One of the top factors contributing to their purchase decisions includes data security (47%), implementing AI and ML (40%), and driving and implementing new technology (40%).
If businesses look at 2019, there is a steady rise in regulations, wherein lawmakers are trying to regulate the new technology. In 2018, the first commercial Deepfakes application was launched. The technology takes a person in an existing image or video and replaces it with someone else’s, using artificial neural networks. Deepfake combines and superimposes the existing media onto source media using machine learning techniques such as autoencoders and Generative Adversarial Networks (GANs). The challenge with Deepfake is that most of the traditional tools weren’t able to identify whether the given clip was fake or real. Deepfake came into news after it was found to be spreading fake news and information about celebrities.
Boston Dynamics company that created Spot, a robot that was viewed across different mediums of technology, gave a true sense of AI development. The four-legged robot was seen doing various tricks such as opening doors, climbing stairs like humans, and even function optimally in wet conditions. The general public had voiced several opinions regarding the robot’s capacity to harm humans. The company then issued a statement stating that the functional capacity of the robot is defined, and it cannot harm any human.
The functional efficiency of robots compared to humans is very low. But do the leaders and businesses understand that, and how exactly are we tackling challenges while making AI. According to many research scientists working in the US, making AI would be easy in the next 10 years, but as humans try to build a technology, it will be as flawed as them.
Let’s talk about the growing disconnect between AI and business,
NeurIPS is a nickname of a conference called a Neural Information Processing System. Over the past 3 years, it has become one of the premier annual events for researchers working on AI. In 2019, the event saw one of the biggest participation with over 13,000 attendees and more than 1,400 research papers being presented. Many of the participants attended meetups that were held in 35 different countries. In the conference, there is an active interest from businesses to find new talent for their R&D department.
Some of the speakers during the conference said that deep learning systems don’t exhibit any human-like learning abilities. They used various examples to showcase that AI is still far from the ability to master new tasks from just a handful of examples, learn new concepts, and using common sense. Researchers even predicted that the field was stumbling in the dark when it came to giving robot qualities like human-efficiency and flexibility. Business leaders present during the conference, however, looked toward building or exceeding human abilities to specific tasks. Researchers, however, had questioned that are businesses setting any benchmarks before bringing an AI system to the level of human decision making.
According to JP Morgan, investment in AI would reach $35.8 billion in 2019. How much of that AI investment is going toward the reskilling of the resources? Currently, only 3% of the business leaders are planning to invest in reskilling significantly in the next 3 years. Most of the organizations are failing to use the available time to ante up their growth as the industry is still building its base.
Some of the business leaders admitted that their primary motivation behind investing in AI was bringing efficiency and reducing the cost of a process rather than promoting growth. According to a survey of business leaders, 76% of the executives said that they had used AI to augment their tasks over the past 3 years, while 90% said that they had used technology to automate tasks.
The flaw that we might see in AI,
Google had recently unveiled a new AI technology called BERT (Bidirectional Encoder Representations from Transformers), which changed the way scientists built systems that learn how people write and talk. BERT was then deployed to various services such as Google’s Internet search engine to predict different search options and results. The technology, however, is still flawed at various levels because it’s just like a child who learns from his parents, both things—good and bad. BERT would even be capable of replicating those flaws at every stage.
BERT is just one of the AI systems, such as Alexa and Google Assistant, which learns a lot from the digitized information available. It considers everything present in the digitized format as knowledge and absorbs it. BERT, for example, is more likely to associate men with computer programming rather than being gender-neutral. Recently, an AI tech was used to predict the news content tone. Everything written in that story was flattering, but, in some places, President Trump’s name was used; and the program gave a negative review about the story.
The rise in AI investments is putting the industry under constant threat of being overly emphasized only on the growth factor. Leaders need to realize that even if they want to develop AI to be effective enough to reduce cost or bring efficiency to the single process, they would eventually want growth from it over time. Building an AI solution business will need insights for both the present and future. To know more about AI solutions, download our latest whitepapers on Artificial Intelligence.