The potential of artificial intelligence is the subject of a new book titled “Embracing the Power of AI”. In the book, five AI experts have come together to take a deep dive into topics like the state of AI today, how organizations can implement it, and why AI will continue to impact our daily lives.
One extract from the book explains how AI is:
Just now reaching its moment of maturity—its tipping point—largely due to recent advancements in machine learning and deep learning, fueled by the enormous amount of data humankind is producing on a daily basis. These developments have emerged as potent drivers of innovation across a number of industries and are powering the rapid acceleration of artificial intelligence investments. As this new version of AI arrives, bringing with it applications that will affect the way that billions of people live and work, it’s important to think beyond the technology and start a new conversation. (“Embracing the Power of AI”, p. 15)
To explore the book’s themes further Digital Journal caught up with one of the authors, J.J. Lopez Murphy. Lopez Murpshy is the Technical Director and Data Science Practice Lead at Globant, an IT and Software Development company.
Digital Journal: How sophisticated is AI becoming?
J.J. Lopez Murphy: AI has become increasingly sophisticated since it was first introduced in commercial applications. In the past, AI has gone through periods called ‘AI winters’, where there would be a sudden drop in activity after the initial hype and investment into AI. But due to recent advancements in machine learning and deep learning, as well as the influx of data available in today’s world, AI has reached a point of maturity.
Now, AI winters are gone for good, and many industries are taking notice. The large amount of data sources and the accessibility to them allow for AI to have the information it needs to grow, and for that growth to be sustainable. AI has also benefited from improvements in processing power and technical prowess, as many different technological tools are able to build off of one another. These conditions have allowed for AI to learn faster and become smarter, making it a viable tool for many more applications than ever thought possible.
DJ: What are the implications from this?
Lopez Murphy: The result from these developments are AI products that have the ability to augment intelligence and add value to the lives of many people (not replace them). Data scientists and researchers have made huge strides in delving deeper into AI to find new ways it can be applied. As a result, businesses are investing in AI and seeing success in its implementation.
Now, a key aspect is that the sophistication comes in two flavours. One is on the technical complexity and the breadth of the problems that can be tackled, and the other one, seemingly opposite, is on the readiness to use and consume by business users. Simplicity is the ultimate sophistication, and any sufficiently advanced technology is indistinguishable from magic.
DJ: How much further can AI develop?
Lopez Murphy: While there is a lot of buzz surrounding AI right now, the technology still has room to grow. New developments and products are continuing to further push the boundaries, so there’s still a long road ahead of uncovering new applications for AI and researching ways to grow the technology. The purpose of AI is to augment the work done by humans and allow people to focus value elsewhere. I think we’ll continue to see more improvements to AI that will further enable this augmentation, and yield greater results for businesses and employees.
DJ: Is there a creative element to this?
Lopez Murphy: Data scientists and companies are also becoming more creative in looking for new uses of AI.
This is best covered by a quote from the book:
In recent months, start-ups and some individuals moved from investigation and research of deep neural networks applied to signal processing (sound, images, and text) to other sorts of jaw-dropping and dazzling applications. Style transferring with deep learning, for example, recomposes an image in the look and feel of another image by using a stream of data as input to apply a certain style to that input to generate an output. So, for example, the input could be an image and the style could be the patterns of Edvard Munch’s painting The Scream. The output will be the original image with the look and feel of The Scream. Another example is text generation using a particular writer’s style. Let’s say you train a model with the whole Shakespeare bibliography and then give some input stimuli to the model; it generates an output text with Shakespeare’s writing style. (Embracing the Power of AI, p. 140-141)
Just as with any technology, there is always more to learn, and scientists are finding plenty of opportunities to experiment with creative uses for AI.
DJ: How close are we to a Turing-like thinking machine?
Lopez Murphy: When it comes to the idea that artificial intelligence will surpass human intelligence, or that robots can ‘think’ as in the case with Alan Turing, I don’t think we are close to reaching this point. The AI we are talking about today is meant to augment intelligence and doesn’t have the emotional intelligence and situational context that humans provide. AI also can’t think on its own like we see in movies, the way Turing predicted. Obviously the technology will continue to grow and improve, but AI won’t advance to a point where it can pose as a human.
Most experts agree though that there is a kind of hard limit on our most successful current approach to AI, namely neural networks, in the sense that while they are extremely efficient pattern detectors, the notion of causality, intent and discovery is still missing. Hence, they learn, but they are not “smart” or “conscious” in the same way we perceive other human beings. We’ll need something else to get there.
DJ: What were the main technological challenges with AI?
Lopez Murphy: The main technological challenges with AI is that the technology is agile and involves experimenting along the way through an iterative process. It’s up to the team to figure out a methodology that works not only for them, but for the AI. Additional challenges that could arise during an AI process are unchecked biases. Since AI learns from what it has been given, teams need to check the data set and algorithms for ethical oversight along the way. The main idea here is to work as a team and understand that AI won’t just do the work for you.
As is discussed in the book:
Expecting to be amazed just because you are calling it AI does not work. Not having a problem to solve does not work. Not knowing what value you are trying to deliver does not work. An ill-defined relationship means that there is not enough trust in the capacity, whether technical or product, to allow for experimentation. It can also be micromanagement that pushes the team toward overcommitting, making promises that cannot be kept. AI projects always have a research bent; thus, they cannot be constrained to a pure “development hours” mind-set. Commitments need to be made and met, and trust must be earned, but through collaboration, not just control. Whether this relates to an external client or the governance team within the project, the outcome is the same. Without trust and commitment, no meaningful advances will be made. (Embracing the Power of AI, p. 99)
AI involves the whole team’s commitment to solve any challenges that might come up throughout the process. In the end, a successful AI project is a team effort.
In a follow-up interview, Lopez Murphy discusses with Digital Journal the impact that AI is having on business and how businesses need to adapt and become more agile in order to remain competitive. See: “How artificial intelligence will disrupt business: Q&A.“
