Mass Layoffs in Tech — Is The AI Winter Coming?

https://miro.medium.com/max/1200/0*g6F1SbnKNZnxam1G

Original Source Here

Until very recently, companies were fighting to attract and retain quality staff in data science. Online business thrived during times of lockdown, with the world suddenly relying on parcel deliveries, cloud environments, online meeting spaces, and virtual pastimes. Tech giants reported record profits, funneling their excess cash into ambitious AI projects and -innovations [1].

Every qualified data scientist was a high-value commodity, and companies bent over backwards to prevent employees from joining the Great Resign movement. Corona or not, the sky seemed the limit for the tech sector.

And then, almost overnight, LinkedIn was suddenly flooded with experienced data scientists looking for another job. Within a matter of days, Twitter fired half of its workforce, Amazon and Meta both cut over 10,000 jobs in mass layoffs, and many more companies either installed hiring freezes or substantially shrunk their work force [2]. Globally, an estimated 200,000 tech workers have lost their job already, and this number will likely rise in the months to come [3].

All of a sudden, it appears the bottom fell out from under the data science community. Are we headed for another AI Winter?

What is an AI Winter?

First of all, what is an AI Winter? Wikipedia [4] defines it as:

“a period of reduced funding and interest in artificial intelligence research.”

The path leading to such a winter is outlined as follows:

“It is a chain reaction that begins with pessimism in the AI community, followed by pessimism in the press, followed by a severe cutback in funding, followed by the end of serious research.”

More broadly speaking, an AI Winter can be classified as a trough in a Gartner hype cycle [5], in which interest in a technology sharply declines when it turns out inflated expectations cannot be met.

Gartner Hype Cycle. If inflated expectations are not met, the technology loses interest and funding is pulled, stalling further developments [image via Wikipedia by by Jeremykemp]

Reportedly, the major AI Winters took place during 1974–1980 and 1987–1993, and people have been predicting another bust will follow sooner or later.

To summarize, for an AI Winter to materialize the following two conditions should be met for an extended period of time:

  • Reduced funding
  • Reduced interest

For the record, empirical evidence for the existence of hype cycles is shaky at best, but we’ll play along for the sake of this article.

Why are data scientists being fired?

Let’s start with the reduced funding. The record layoffs of people in tech companies naturally decrease the capacity to further develop AI.

Obviously, not all people fired are data scientists, and not all data scientists design AI. Nonetheless, most people in tech roles do use AI in their daily work, one way or another.

In more applied roles, you might not even notice innovations directly. However, in the long run, consider what happens without inventions to multiply matrices more efficiently, quicker computations of gradients, practices to transparently explain automated decision-making… How effective would you be with the toolkits of five years ago?

When these kinds of innovations stall, the sector as a whole will stagnate, and data scientists will be less impactful than they could be. AI is so intertwined with the many branches of data science, that the effects of the mass layoffs will trickle through all crevices of the domain. Naturally the unfortunate ones who actually lost their jobs are impacted most, yet all of us will be affected by a loss of AI innovation power.

From a common sense business perspective, the reasons for the layoffs are quite straightforward though:

  • High costs reductions: Data science is known for its high wages and substantial bonuses; it’s one of the reasons so many people try to break into the field. Consequently, the cuts have a substantial and direct impact on the operational costs of companies.
  • Deprioritizing R&D: Although the concept of ‘data science’ is rather broad, many in the field are involved in research & development in some way. In times of crisis, R&D activities always take hits, with the focus being on short-term survival rather than long-term visions and speculative endeavors.
  • Correcting underperformance: Tech stocks have experienced big falls in recent times. It appeared that corona would drive permanent changes towards an ever-expanding digital universe, and the tech sector expanded accordingly. However, realized performance does not match the rose-tinted expectations.

Some concrete examples?

  • Meta sank billions into the Metaverse — losing nearly 10 billion on the project this year alone [6] — with no break-even point in sight yet.
  • According to Musk, Twitter is currently losing $4M a day [7].
  • Amazon recently became the first company in history to lose one trillion (!) in market value, with Microsoft trailing not much behind [8].
  • Google continues to experience shrinking profits, partially due to an oversaturated ad market and partially due to failed innovations [9].

On a more granular level, specific teams or products fail to yield profits, regardless the qualities of the members or the brilliance of the idea. More on that later.

In the end, layoff decisions are often simply a question of how much a team costs and how much it generates. There is office politics and business visions, but the bottom line ultimately matters.

Photo by Maxim Hopman on Unsplash

AI Winter or ‘just’ another crisis?

The (pending) reduction in funding for AI is undeniable, but at surface level, there are obvious macro-economic reasons for the layoffs. The global economy recovered surprisingly quick and well from the corona crisis — in part due to near-unlimited funding from governmental bodies — but the war in Ukraine triggered another cascade of problems, including further supply chain disruptions and soaring energy prices. Inflation rates went through the roof, consumers had spending power, people grew fearful… That’s all the ingredients a crisis needs.

Economic headwind and layoffs go hand-in-hand, so trimming down on staff costs alone is not enough to constitute an AI Winter. However, if we take a closer look to who were fired, we may perceive recent developments as more than bracing for the storm. Time to consider some examples:

  • The dissolution of Twitter’s entire Ethical AI Team garnered wide-spread attention, as the team was considered leading in the thrust towards transparent and unbiased AI [10]. The cut might be interpreted as an act in a one-man show, yet similar targeted layoffs might be seen in other tech companies as well.
  • Meta’s Probability Team, working on topics such as probabilistic- and differentiable programming that could aid ML engineers, was dissolved entirely. Reportedly, it was a world-class team of experts, but seemingly it lacked a sufficiently visible impact [11].
  • Amazon reportedly fired large parts of its robotics- and devices divisions, marking a reorientation towards services proven to generate cash flows [12,13,14].

In these decisions, it should be considered that tech giants — while obviously not philanthropists — have mountains of cash at their disposal. As such, pulling the plug on AI projects is not essential to short-term survival, it means they lost faith in their profitability or value in the longer run.

Terminating projects occurs at all times, but at the moment a lot of plugs are being pulled. For various companies it is the largest staff reduction in decades; it is hard to overstate the magnitude of present events.

Being in the middle of the process and lacking comprehensive statements on the size and scope of the restructuring efforts, it is still too soon to see in what direction AI will move. However, given that even world-class AI experts are no longer guaranteed a job, it appears there is more at play than simply anticipating economic setbacks.

Photo by Carles Rabada on Unsplash

What is next?

How the future pans out will evidently depend on many factors: the war, the energy crisis, the success of anti-inflation measures, sentiment among consumers, etc. Still, a V-shaped recovery (a rapid implosion followed by an equally quick rebound) as experienced during corona seems unlikely. A U-shaped pattern (gradual decline, stagnation, slow recovery) seems to be the best we can hope for [15]. Given the sizeable reductions in the tech workforce, it will take substantial time before we are back at the levels we started 2022 with.

Does all of this imply a looming AI Winter? The reduction in funding and manpower seems to be a given, and the targeted eliminations and slimdowns of many AI divisions definitely can be interpreted as a reduced interest in AI, or at least branches of the field.

Having that said, AI development will certainly not stop. Even previous winters never halted AI progress completely. Besides, the last winter occurred in early the 90s. Present-day AI is so sizeable and so deeply ingrained in everyday life, it is hard to imagine a real ‘break’ in AI developments.

Although the massive layoffs, the termination of many AI initiatives and the present short-term focus of companies are unlikely not to hurt the progress of AI, the economic headwind appears to be a much stronger driver than a loss of faith in AI in general. As such, a severe AI Winter is not likely — Artificial Intelligence simply has too much going for it still.

That said, an extra blanket might not hurt in the times ahead of us.

Photo by Robert Thiemann on Unsplash

AI/ML

Trending AI/ML Article Identified & Digested via Granola by Ramsey Elbasheer; a Machine-Driven RSS Bot

%d bloggers like this: