Were you unable to attend Transform 2022? Check out all of the summit sessions in our on-demand library now! Watch here.
A recent New York Times article concludes that new AI-powered automation tools such as Codex for software developers will not eliminate jobs but simply be a welcome aid to augment programmer productivity. This is consistent with the argument we’re increasingly hearing that people and AI have different strengths and there will be appropriate roles for each.
As discussed in a Harvard Business Review story: “AI-based machines are fast, more accurate, and consistently rational, but they aren’t intuitive, emotional, or culturally sensitive.” The belief is that “AI plus humans” is something of a centaur, greater than either one operating alone.
This idea of humans plus AI producing better outcomes has become a tenant of faith in technology. Everyone talks about humans being freed up to perform higher-level functions, but no one seems to know just what those high-level functions are, how they translate into real work and jobs, or the number of people needed to perform them.
A corollary of this augmented-workforce narrative is that not only will AI-augmented work enable people to pursue a higher level of abstract thinking, it will — according to some — also lift all of society to a higher standard of living. This is certainly an optimistic vision, and we can hope for that. However, this could also be a story imbued with magical thinking, with the true end-game being fully automated work.
MetaBeat will bring together thought leaders to give guidance on how metaverse technology will transform the way all industries communicate and do business on October 4 in San Francisco, CA.
What does the evidence tell us?
Don’t get me wrong; there is some evidence to support the view that AI will help us work rather than take our jobs. For example, AI lab DeepMind is designing new chess systems for the two intelligences to work in tandem with humans rather than opposed to them.
And Kai-Fu Lee, the Oracle of AI, also buys into this promise. In his new book, AI 2041: Ten Visions for our Future, he argues that repetitive tasks from stacking shelves to crunching data will be done by machines, freeing workers for more creative tasks. Forrester Research has likewise articulated that AI deployment enables people to better use their creative skills.
But, of course, some people are more creative than others, meaning that not everyone would benefit from AI-augmented work to the same degree. Which in turn reinforces a concern that AI-fueled automation, even in its augmented work capacity, could widen already existing income disparities.
One problem with the AI-augmented workforce promise is that it tells us AI will only take on the repetitive work we don’t want to do. But not all work being outsourced to AI is routine or boring.
Look no further than the role of the semiconductor chip architect. This is a highly sophisticated profession, an advanced application of electrical engineering in arguably one of the most complex industries. If ever there was a job that might be thought of as immune from AI, this would have been a strong candidate. Yet recent advances from Google and Synopsys (among others using reinforcement learning neural network software) have shown the ability to do in hours what often required a team of engineers months to achieve.
One ever-faithful tech watcher still argued that the algorithms will “optimize and accelerate time-intensive parts of the design process so that designers can focus on making crucial calls that require higher-level decision making.”
A step along the path to more complete automation
More than likely, the current perception of work augmented by AI is a reflection on the current state of the technology and not an accurate view of the future when automation will be far more advanced. We first saw the potential of neural networks a decade ago, for example, and it took several years until that technology was developed to the point where it had practical advantages for consumers and business. Fueled in part by the pandemic, AI tech is now being widely implemented. Even massage therapists should take note, as a robot masseuse can now deliver a deep tissue massage. Yet, these are still early days for AI.
AI advances are being led by improvements in both hardware and software. The hardware side is driven by Moore’s Law, the idea that semiconductors improve by roughly 2x the number of transistors – producing roughly equivalent performance and power efficiency gains – every couple of years (and similarly drive down the costs of computing). This principle has been credited with all manner of electronic advances over the last several decades. As noted in a recent IEEE Spectrum article: “The impact of Moore’s Law on modern life can’t be overstated. We can’t take a plane ride, make a call, or even turn on our dishwashers without encountering its effects. Without it, we would not have found the Higgs boson or created the Internet.” Or have a supercomputer in your purse or pocket.
There are reasons to think that Moore’s-Law driven improvements in computing are nearing an end. But advanced engineering, ranging from “chiplets” to 3D chip packaging promise to keep the gains coming, at least for a while. These and other semiconductor design improvements have led one chip manufacturer to promise a 1000x performance improvement by 2025!
The expected improvements in AI software may be equally impressive. GPT-3, the third iteration of Generative Pre-trained Transformer from OpenAI, is a neural network model consisting of 175 billion parameters. The system has proven capable of generating coherent prose from a text prompt. This is what it was designed to do, but it turns out that it can also generate other forms of text as well, including computer code and can also generate images. Moreover, while the belief is that AI will help people to be more creative, it could be that it is already capable of creativity on its own.
At its launch in May 2020, GPT-3 was the largest neural network ever introduced, and it remains among the largest dense neural nets, exceeded only by Wu Dao 2.0 in China. (At 1.75 trillion parameters, Wu Dao 2.0 is another GPT-like language model and probably the most powerful neural network yet created.)
Some expectations are for GPT-4 to also grow and contain up to a trillion parameters. However, OpenAI CEO Sam Altman has said that it will not be larger than GPT-3 but will be far more efficient through enhanced data algorithms and fine tuning. Altman also alluded to a future GPT-5. The point being that neural networks have a long way to run in size and sophistication. We are indeed in the midst of an age of AI acceleration.
In the new book, Rule of the Robots: How Artificial Intelligence Will Transform Everything, author Martin Ford notes that “nearly every technology startup is now, to some degree, investing in AI, and companies large and small in other industries are beginning to deploy the technology.” The pace of innovation will only continue to accelerate as capital continues to pour into AI development. Clearly, whatever we are seeing now in the way of AI-powered automation, including the belief that AI will help us work rather than take our jobs, is but an early stage for whatever is still to come. As for what is coming, that remains the realm of speculative fiction.
In Burn In: A Novel of the Real Robotic Revolution, a Yale-educated lawyer is among those impacted when his firm replaced 80% of the legal staff with machine learning software. This could happen in the near future. The remaining 20% were indeed augmented by the AI, but the 80% had to find other work. In his case, he winds up doing gig work as an online personal assistant to the wealthy. Currently, startup company Yohana is working to realize a variation of this vision. The company is initially offering a blend of human and AI services, starting with a living, breathing assistant that draws on data to tackle the to-do lists of subscribers. It will be telling to see if these assistants will be like the secretaries of yore, but wielding AI, or if they will be displaced cognitive workers.
The AI-driven transition to a largely automated world will take time, perhaps a few decades. This will bring many changes, with some being highly disruptive. Adjustments will not be easy. It is tempting to think that ultimately this will enrich the quality of human life. After all, as Aristotle said: “When looms weave by themselves, man’s slavery will end.” But embracing the AI augmented work concept as currently articulated could blind us to the potential risks of job loss. Kate Crawford, a scholar focused on the social and political implications of technology, believes AI is the most profound story of our time and “a lot of people are sleepwalking into it.”
We all need to have a clear-eyed understanding of the growing potential for disruption and to prepare as best we can, largely by acquiring those skills most likely to be needed in the coming era. Companies need to do their part in providing skills training, and retraining will increasingly need to be a near continuous process as the pace of technology change accelerates. Government needs to develop public policies that direct the market forces driving automation towards positive outcomes for all, even while preparing for a growing social safety net that could include universal basic income.
Gary Grossman is the Senior VP of Technology Practice at Edelman and Global Lead of the Edelman AI Center of Excellence.
Welcome to the VentureBeat community!
DataDecisionMakers is where experts, including the technical people doing data work, can share data-related insights and innovation.
If you want to read about cutting-edge ideas and up-to-date information, best practices, and the future of data and data tech, join us at DataDecisionMakers.
You might even consider contributing an article of your own!