AI, Automation, and the Future of Workplaces
Introduction
Earlier this year, I came across a quote by Einstein which said:
“It has become appallingly obvious that our technology has exceeded our humanity.”
This quote brought to mind the ambivalent conversations we witness on the nature of modern AI each day. It begs the question as to whether or not we can truly quantify the impact of this Pandora’s box that lies in front of us. A good place to begin would be to make the distinction between what can be classified as “AI”.
Now, imagine less Arnold Schwarzenegger Terminator and more along the lines of a smart toaster that knows just the way you like your morning order. At its simplest, AI comprises a combination of computer science and robust data sets which enable problem-solving. Through methods such as reinforced learning, subfields of AI such as Machine Learning help in crafting expert systems that can make predictions based on input data.
This is just one instance among countless others where AI truly comes into its own: cutting down the drudgery involved in many work tasks. The repetitive and tedious nature of manual labor lends itself room for human error. The human mind can only comprehend and carry out instructions for such a limited period before even the most repeated acts are susceptible to errors due to the nuances that seep in as a consequence of fatigue. Automation is perfectly suited for such standardized processing work which helps free up human operatives. This in turn opens up numerous other channels to concentrate human effort on more interpersonal and creative aspects in work, as well as in our personal lives.
The Promise of Emerging Tech
By understanding the scope of the technology we harness, can we truly begin to break down its impact on the current and future markets. Sectors such as Banking, Automobile Production, Weather Forecasting, Advertising, and many others have seen an instant impact by ushering in the transition to more AI-centric systems. Doing so has reduced operational and employee costs substantially. From the inception of Henry Ford’s assembly line first introduced in the West in the 1920s, the idea of cost-cutting was placed at the forefront. Since then, auto manufacturers have steadily automated the assembly process with robotics, with humans functioning in primarily supervisory, technical, and quality assurance roles.
When new technologies make bold promises, we must discern the hype from what’s commercially viable. Gartner Hype Cycles come into play by providing a graphic representation of the various stages of maturity and adoption of tech and applications. It assesses the feasibility and potential relevance of emerging trends to solve real business problems to show how an application will evolve over time.
Photo: A Gartner Cycle depicting the various stages
Here’s how the Hype Cycle defines the five key phases in the technology’s life cycle:
· Innovation Trigger: The launch of a breakthrough innovation via a public demonstration. Mostly comprises proof-of-concept stories to trigger media publicity.
· Peak of Inflated Expectation: A “buzz” is generated which is attributed to early success stories which garner expectations that overshoot its current capabilities.
· Trough of Disillusionment: Unsatisfactory results lead to concerns and skepticism regarding potential value. Issues in performance and adoption speeds lead to unfulfilled expectations, hence disillusionment sets in.
· Slope of Enlightenment: The early backers overcome the initial setbacks and gain a competitive advantage. Use cases for the innovation are further brought to light, in addition to whether or not they can generate value.
· Plateau of Productivity: After successfully demonstrating the benefits, the technology is adopted by the broad market. Organizations feel comfortable integrating it within their ecosystem with considerably little risk.
The incorporation of AI and ML by flourishing tech giants such as Amazon (AWS), Google (Google Cloud Platform), and IBM has demonstrated the capabilities of having a business powered by the most emergent phase of innovation seen in the past century.
‘It’s Pizza Time’: The Positive Side of AI
Photo: Domino’s Innovation Garage
In 2019, Dominos Ferny Grove set a new record for fast food — consistently delivering freshly-made pizzas to customers’ houses in less than six minutes for an entire week.
‘Slow where it matters, fast where it counts’, is what Domino’s Australia CEO Nick Knight had to say as part of the company’s project 3TEN in association with AWS.
They achieved this feat by:
· Creating a data lake consisting of key order information for data storage with the help of Amazon Simple Storage Service(Amazon S3).
· Using AWS Glue for data querying
· Using Amazon SageMaker to build and train machine learning models to predict the likelihood that an order will be placed.
The employees of the store could see the ordering list displayed on the screen. Specific pizzas with various color indicators corresponding to the likelihood of those pizzas being ordered. The aim was to have pizza ready for pick up within 3 minutes and safely delivered within 10 minutes.
The food chain partnered up with Dragontail Systems to craft a quality control system, where food safety and hygiene are paramount. With a robust delivery dispatching system that can automatically detect which drivers should take orders, delivery routes are optimized without getting the manager involved.
The above example marvelously encapsulates the ethos behind AI. Every business has a unique and urgent pain point that needs to be addressed. Once ‘trained’, a system can comprehend customer and business standards and alert the manager when these standards have not been fulfilled.
We may witness a rise in AI and robots working in tandem with human operatives. These so-called “cobots” complement, rather than fully replace their human counterparts and improve overall efficiency.
AI and its applications could help us tackle societal moonshot challenges in areas ranging from material science to medical research and climate science.
By providing real-time data on disasters and weather events, the risk of human fatality is considerably reduced. Deep learning will soon be integrated with disaster simulations to come up with response strategies.
‘Another One Bites The Dust’: Concerns Surrounding AI
1. Pitfalls Faced While Defining ‘End Goals’:
The lines of code that make our machines tinker will inevitably lack nuance. This brings the morality of AI into question whose goals and incentives might not align with our true preferences.
Photo: Garry Kasparov plays against IBM Supercomputer ‘Deep Blue’, circa 1997
Today’s AI is a success at accomplishing tasks with a definite end-goal such as Jeopardy! and Go. The problem lies when we ask a free-roaming, “autonomous” robot to optimize a “reward function”. Since it is impossible to include and correctly weigh all goals, subgoals, and caveats in the reward function, we’re left with misaligned AI. The significant risks are felt when robots will be ruthless in pursuit of their reward function at the expense of other parameters which accumulate in damages.
2. Automation Will Lead To Displacement Of Workers:
A study by McKinsey analyzed more than 2000 work activities over 800 occupations to determine which categories of work were more easily automatable than others. Roughly half of the activities that people do across all sectors are based around highly predictable and structured environments and include data collection and processing. It was concluded that about 30–60% of such activities could be fully automated which could affect over 400–800 million workers as soon as the year 2030. A shift in required workforce skills may be seen. Demand for advanced technological skills such as programming will grow rapidly. This will put additional pressure on the already existing workforce-skills challenge. One lesson of the past decade is that while globalization may have benefited economic growth and consumerist tendencies in people, wage and dislocation effects on workers were not adequately addressed. The risk is that automation could exacerbate wage polarization and income inequality, primarily affecting middle-wage jobs.
This isn’t to say those at the top are safe from the perils of becoming obsolete. It is a constant task of a developer to automate a significant portion of his current work so that they may take on more tomorrow. With the emergence of AI tools like GPT-3, DALL-E 2, Google Imagen, and more, AI can now accept a command in plain English and respond intelligently in return.
This would prove to be the pinnacle of declarative programming: a method to abstract away the control flow for logic and instead, state what the task or desired outcome is. GitHub Copilot partially achieves this by offering smart suggestions for problems written in plain-English. Although limited in its capacity at present, it is presumptuous to see this technology, understand it and where it came from without being able to see where it’s headed tomorrow. Unbounded growth is the nature of AI.
Photo: A demonstration of DALL-E 2’s capabilities in generating an image from text using Natural Language Processing(NLP)
Conclusion:
“Deus Ex Machina” — A God From The Machine
The Latin phrase describes a plot device, where a seemingly unsolvable problem in a story is suddenly resolved by an unexpected event. In the context of humanity, it isn’t hard to imagine that within a few decades or centuries, survival of the fittest would clear the way for technological selection rather than through natural selection.
The landscape of a future driven by AI brings great promise, one which we must harness and channelize our efforts into. There is work for everyone today and there will be work for everyone tomorrow, even in a future with automation.
