2021: The Redemption Arc

It is about this time each year that I like to take a moment to reflect on the prior twelve months in technology and machine learning; and, of course, to make some semi-educated prognostications about the forthcoming twelve months. And I will still do exactly that; but I’d like to take an additional moment to first reflect more broadly on the chaos that was 2020.

In March (which now feels like eons past), I wrote some thoughts on the sudden, and likely to be jarring, transition into remote work that a lot of companies were about to endure. Thankfully, Aiera has been a remote organization since inception, which has certainly helped us handle the pandemic as a company; but that origin can do little to prepare us, as individuals, for what was to become life outside of work.

In the early days and weeks of the first phases of quarantine, when infections were starting to spike and protocols were just being formed and followed, I spent quite a few hours trying to find a new normal. I’m already a bit of a workaholic, so filling most of the day (and a chunk of the night) was not a particular challenge; but in what hours I did typically reserve for myself, I tried to fill that time with new home-bound efforts, as did most of us. At first I trialed the more common things, such as increasingly complicated baking recipes, learning to play the guitar, reading more frequently, and going for a lot of long runs (to get some fresh air). I bought fanciful masks, stocked up on toilet paper and non-perishables, grew a beard, let my hair grow long and wild, and put on the occasional stunt to try to keep my team and my family amused (such as attending an all-hands meeting in formal wear).

While the work of Aiera carried on mostly unchanged, thanks in part to a financial market that proved remarkably resilient to pandemic impacts, the weight of the world, both personal and otherwise, crept in. Watching infections (and deaths) rise was terrifying; in those early days, when MA and NY were first being trampled by the virus, a cloud hung over everything. The world was in dire straits, and reading the news became a desperate chore. Here and there, unexpectedly, the impact would suddenly come closer to home: a few relatives would contract the virus and became ill; friends would lose their job; and a few beloved stores and restaurants would close their doors for good…

Personally, the reality of increasing and persistent isolation became hard to reconcile; visiting family and friends, spending any day or night out of the house, and even the less obvious activities, like in-person meetings, business travel, conferences, and shopping; all evaporated. The world suddenly shrank dramatically. Previously mundane but oddly pleasurable tasks, like grocery shopping, became carefully engineered, rigid activities, devoid of spontaneity or calm. This was all essential, of course, but still a taxing change to reality.

After a few weeks of rather random efforts to fill the gaps, I started taking a few hours in the early morning to head up into the hills and hike; quiet and alone in the woods. At first, this was little more than an effort to get some exercise. But after a few days, I found that a little touch of nature and fresh air was just enough to give me that needed bit of clarity and focus. It was something of non-work purpose to enjoy each morning (along with that bit of exercise).

It was while hiking that I often came upon little glimmers of hope as the year unfolded. Whether those glimmers were personal, such as listening to song-birds fill the air on a sunny Saturday morning; or worldly, such as getting that first notification that moderna, a Cambridge-based company (yay Massachusetts!), had crafted an mRNA vaccine candidate and entered phase one trials. Each little glimmer helped remind me that we humans, as stubborn and awful as we can be at times, are ultimately capable of weathering and solving any challenge. As vaccinations, including the moderna creation, are being approved and administered at scale, less than a year after first entering development (in an unparalleled scientific achievement worthy of parades and folk songs), I marvel at what we’re able to do when we’re united and dedicated to a single task.

All evidence to the contrary, I have always and will always be a humanist at heart. When we really need to do something, when the chips are down, when our backs are against the wall, humans will always figure it out…

And with that optimistic thought in mind, let’s get back to the core premise of this end-of-year retrospective and focus on what notable achievements were seen in 2020, and talk a little bit about what further achievements I expect to see in a (hopefully) vastly improved 2021.

Linguistics

No discussion of 2020 technical achievements would be complete without discussing the OpenAI marvel that is gpt-3. First released for limited public access in June, following a May publication on the underlying technology, this language model has quickly shaken the NLP landscape and captivated the public attention. Trained on hundreds of billions of words that have been collected from various data sets from across the public web, it is by far the largest language model trained to date. I was lucky enough to gain access to the API in the fall, and have been incredibly impressed at the scope and capability of this 175 billion parameter generative model.

In September of this year, Microsoft negotiated exclusive access to the underlying source code for building the model, while preserving API access for those granted. Engineers have so far used gpt-3 to create fast and powerful search engines, code generators, and chatbots; but the full power and capacity of this innovation is likely not to be seen for some time.

The world of computational linguistics has been in the spotlight for a number of years. Text and audio-based interfaces, including Alexa and Siri, have been blazing a path since the early 2010’s; start-ups in law, academia, healthcare, finance (shameless plug), and beyond have been launched to provide analysis and generative insights across a variety of language sources. Even the security industry has been using NLP to suss out bots and other malevolent behavior on social media. In the coming year, NLP will continue to drive significant advances across the technology spectrum, and across all industry verticals.

Autonomous Driving

This has been a hot-button topic in the world of machine learning for several years, and one which frequently ebbs and flows between unbridled optimism and despondent obituary. But in October, Waymo, the autonomous driving unit of Google that has raised billions of dollars in the last decade, announced that fully-driverless taxis (without even a safety observer onboard) were now available for public use on the streets of Phoenix!

While not quite the worldwide revolutionary bellwether that everyone has been awaiting for some time, this focused release was a major milestone in the inevitable shift toward increasingly automated transportation solutions. As some efforts fail, such as Uber (which sold it’s autonomous efforts this year) and others achieve incremental success, such as Tesla (which has begun rolling out Autopilot for in-city use), autonomous driving is likely to make continued, steady gains in 2021. Expect the rollout of Waymo to hit more cities and become more of a media draw over the course of the year.

Healthcare

Elon Musk is unquestionably a little out there, and a loved one should probably get him to stop tweeting. But it’s hard to overlook the announcement he made in August that one of his many ventures, Neuralink, had inserted neural implants into the brain of a pig, named Gertrude, and were monitoring and recording neural activity in great detail. The applications of this kind of technology are a hot debate amongst technologists and neurologists, and Neuralink has been famously secretive about their ambitions. But this simple demo has provided ample fodder for future thinking.

And, to co-exist along the shining healthcare achievement of 2020 (which is, of course, the multitude of COVID19 vaccines that have been developed and deployed) are numerous technical efforts in healthcare to mitigate the damage of the pandemic. These include contact tracing operations (natively available in iOS), telemedicine efforts, and digital psychology to head off the mental health risks of isolation. In a continuance of ongoing efforts, expect further advances from machine learning and technology in healthcare, including diagnostics, and telemedicine and at-home care.

Internet

While we’re hovering around the topic of Elon Musk and his various ventures, it’s also worth mentioning the initial rollout of Starlink, the worldwide satellite internet service set to be offered by SpaceX. Just shy of a thousand satellites were launched in 2020, and the first beta users have been provided access across the northern regions of North America. Musk has ambitions to provide low-cost access to low-latency satellite internet around the world, and it seems entirely possible that this might be achieved in the next few years.

While technology and software have eaten the world, it is often overlooked that internet access is not a universality. By estimates, just 4.3 billion people have any access to the internet; that leaves over 3.5 billion people without any connectivity. And, it’s worth noting, access amongst the connected is not of consistent quality, especially in rural locales. Offerings like Starlink (and similar efforts from Facebook, Google, and Blue Origin) are likely to have a revolutionary effect on rural and disconnected populations, and will dramatically expand the scope of internet businesses.

Cybersecurity

Following the 2016 Presidential election, the pitfalls of social media manipulation became an important and frequent area of discussion. But the issue of nation-state manipulation, and cybersecurity in general, reached a fever pitch toward the end of 2020 as a massive, multi-part data breach was announced in December. Frequently cited as the worst cybersecurity breach in history, vast swaths of American government were left either vulnerable or penetrated, including the Department of Defense, State, Commerce, and Energy, including (alarmingly) parts of the nuclear security administration.

2021 is likely to become a major year for cybersecurity. In addition to the aforementioned security breach, the issue of disinformation will continue to exert considerable influence on the industry, and technical solutions to these problems will remain highly sought after. With the advent of a new Presidential administration, expect this to get renewed national focus.

Travel, Retail, and Restaurants

On-the-ground realities for these industries are dire. Restaurants are either closed or heavily restricted; travel remains highly depressed, even through the holidays; and physical retail outfits continue to struggle (even as online retail, such as Amazon, has flourished in this new at-home paradigm). Coupled with these pains, supporting technologies have seen either a collapse in business (as is the case with many travel tech companies) or a major shift in demand and priorities. But assuming the vaccines that are currently being deployed at increasing scale do prove highly effective, there could be a massive resurgence of demand in many of these impacted industries. This will include an equal resurgence in the technical support that surrounds them.

Additionally, this chaos has drawn attention to some constraints and bottlenecks that had existed within our system that will likely need to be addressed. For example, during the online retail boom seen early in the outbreak, Amazon encountered such a massive surge in orders that extended delays became commonplace, and most shipping outlets became overwhelmed. Addressing this logistical fragility will likely include heightened focus on autonomous efforts and vendor diversification.

Expect well-funded new entrants into each of these verticals, as the vulnerable incumbents try to regain their footing once the world starts to, carefully, look a little more normal heading into the summer or fall.

Quantum Computing

In what has become a perennial mainstay of my retrospectives, quantum computing continues to make careful, minuscule strides into actual reality. In October of this year, Google published a paper claiming that it had achieved “quantum supremacy,” which means it performed a calculation using their quantum computers that is not achievable on “classical” hardware.

Google and IBM are locked in something of a quantum feud, with IBM offering limited public access to quantum machinery, and promising a significant scale in quantum hardware in the next year. While this continues to exist mostly on the fringe of mainstream technology, I continue to expect an increase in attention and achievements in the next year. Once hardware becomes even lightly commoditized, research and advancement in the space will explode, especially in the security and financial world.

A Final Thought

Separate from technology, there is a wider question that is likely to be answered at some point in 2021: is this remote revolution truly a revolution, or is it a temporary necessity of the times. Or, more likely, is the future reality something of a middle ground, with hybrid solutions becoming the norm.

There is no question that the current environment is not a true and proper experiment in remote work. The world is on fire, so any measured gains or losses in productivity must account for the stresses and other intangibles of working through a global pandemic. Nonetheless, this is still an experiment of some kind, and the burning question remains: once the world is vaccinated, and the worst of the pandemic has come to a close, will companies reverse course and bring employees back to the office? Or is this remote shift a permanent change? Assuming vaccine success, we may start to get some answers at some point in the latter half of 2021…