Big Tech’s Layoffs, AI, and the Closing of the Productivity Gap

Big Tech has let go of thousands of workers in the last couple of months. In addition to the end of the era of cheap money and a broader economic slowdown, this story may have another angle.

This is the impact of AI and the possible closing of the “Productivity Gap.” 

The Productivity Gap is a phenomenon where workers’ output, especially in developing countries, has been growing slower than expected. The shift to cloud computing and SaaS business models in the mid-2010s led to an explosion in both the valuations of technology companies and increases in the productivity of individual engineers and teams. A small startup could spin up and scale a business faster than ever. 

Fast forward to the mid-2020s, and suddenly cloud computing is a commodity. Innovative Frameworks from the last decade, like React, Spring, and others, are bloated and complex. 

For the last few years, companies like Meta, Alphabet, and Microsoft could hedge their bets and grow their teams because they were less likely to become disrupted by a small startup. Hoarding talent and doing “acqui-hires” was a feasible strategy.

Explaining the Tech Layoffs

Now there is once more a disruptive technology on the horizon. Generative AI Models are making giant leaps – a small team of ML-native programmers could build something that could blow incumbent services out of the water. 

Alphabet’s panic over OpenAI’s ChatGPT is a case in point. Suddenly it doesn’t make sense to hoard talent to work on a platform that is about to be irrelevant. 

AI-enabled software and infrastructure could close the productivity gap and fuel the rise of disruptive startups. 

The incumbents are then cutting costs and preparing themselves for the next round of disruption by making significant investments in AI. 

It no longer makes sense to hoard programmers when the entire industry could undergo a paradigm shift similar to that brought about by Cloud Computing 15 years ago.

The brutal layoffs we have seen in the last three months could be the result.

The Limits of Generative AI

AI is having a moment. The emergence of Generative AI models showcased by ChatGPT, DALL-E, and others has caused much excitement and angst. 

Will the children on ChatGPT take our jobs? 

Will code generation tools like Github Copilot built on top of Large Language Models make software engineers as redundant as Telegraph Operators? 

As we navigate this brave new world of AI, prompt engineering, and breathless hype, it is worth looking at these AI models’ capabilities and how they function. 

Models like the ones ChatGPT uses are trained on massive amounts of data to act as prediction machines. 

I.e., they can predict that “Apple” is more likely than “Astronaut” to occur in a sentence starting with: “I ate an.. “.

The only thing these models know is what is in their training data. 

For example, GitHub Copilot will generate better Python or Java code than Haskell. 

Why? Because there is way less open-source code available in Haskell than in Python. 

If you ask ChatGPT to create the plot of a science fiction film involving AI, it defaults to the most predictable template. 

“Rogue AI is bent on world domination until a group of plucky misfit scientists and tough soldiers stops it.” 

Not quite HAL9000 or Marvin the Paranoid Android. 

Why? Because this is the most common science fiction film plot.

Cats and Hats

Generative AI may generate infinite variations of a cat wearing a hat, but it has yet to be Dr. Suess. 

AI is not going to make knowledge work obsolete. But, the focus will shift from Knowledge to Creativity and Problem-Solving. 

Climate Change and Category Errors

Stuart Kirk, former journalist for the FT and now former banker at HSBC got into trouble last week for suggesting that climate change risks are overblown.

Before his suspension, he was the head of Responsible Investing for HSBC asset management.

Here is his presentation. It’s worth a watch.

In his presentation he says:

  • Climate change risks are overblown
  • The “markets”, in all probability, already priced in climate change risks
  • Climate change adaptation is more pragmatic and likely cheaper than mitigation
  • By the time climate change hits, we will all be dead anyway. So why bother?

I found the presentation interesting and a little horrifying – in the drunk uncle holding forth – sense. He makes some good points – about the short term nature of markets and investing, about the necessity of climate change mitigation, for example. But the general attitude can be summarised as 🥱🤷‍♂️.

I am still surprised that after 2008, after COVID, Ukraine and all the other shocks, people like Mr Kirk still think in terms of normal distributions. I.e. the probability of events can be modelled as a bell curve – with very bad or very good events having low probabilities, and predictable “average” events being the most common.

Or to channel mathematician, philosopher and truculent Twitter warrior N. Taleb, the likes of Mr Kirk believe that the impact of climate change to be an ergodic process while it is most definitely not.

Doing a Google search for “Ergodicity” will lead you to baffling mathematical and statistical explanations. But it is, at its core, an intuitive concept. In a non-ergodic system, things that are true for the aggregate may not be true for the individual.

In Mr Kirk’s presentation he plots economic growth from the 1930s to the present day and states, pretty much, that the “line goes up” despite world wars, economic upheaval, recessions etc. He uses this trend to assert that we will be fine despite the risks of climate change. The benefits of a growing economy will overcome the downsides of climate change.

However, the story of aggregate growth over the last 100 years hides tales of individual ruin.

For example, someone who invested all their savings in tech stocks in 2002 probably didn’t have anything left to make money when the market finally moved up. For those unlucky investors, it was game over. Therefore, we are modelling a process that is non-ergodic (individual outcomes can be radically different than aggregate outcomes) as an ergodic process.

So, what does this have to do with climate change?

I believe that the effects of climate change make our economic system even more non-ergodic. It makes it much more likely to have extreme events – heat waves, wild fires, hurricanes, droughts, etc. This makes modelling based on aggregate probabilities a little suspect. Sure, you could increase insurance premiums for coastal communities to account for higher flooding risk. This is what Mr Kirk means by the risk being “priced in”. But what happens when entire communities are wiped out due to an unprecedented storm surge, or heat wave, or forest fire?

Climate change adds more chaos to a complex system. It heightens the likelihood of extreme events that have catastrophic outcomes. Adaptation measures are necessary but they will do little to mitigate the impact of “black swan” events. So it doesn’t matter how complex your modelling is, and how sophisticated your investment strategy is. If you die due to a freak hurricane, you are done.

The likes of Mr Kirk are making a category error. The only way to “win” in an non-ergodic system is to survive. We should be thinking of what can be done to ensure that we don’t face catastrophic loss, so that we can continue to reap the benefits of growth in the future.

Further Reading

• An excellent primer on Ergodicity

Nassim Nicholas Taleb on Ergodicity

Elon Musk & The Twitter Algorithm

I have been trying to avoid the whole Elon Musk / Twitter drama, but it has been challenging. I am ambivalent about whether Mr. Musk’s takeover of Twitter is a good or bad thing. My vibe is 🤷🏾‍♂️.

But, I do have an issue with one of Mr. Musk’s ideas: open-sourcing the Twitter algorithm to ensure there is no “bias.”

I think this is disingenuous, and Mr. Musk is playing to his (adoring) audience a little bit. 

It is improbable that there is the “one true algorithm” at Twitter. They probably use a combination of machine learning-based recommendation models with other systems such as entity and intent detection. Take a look at Twitter’s engineering blog to see how much ML drives recommendations on the social network.

So, if the intention is to look at the code and delete any (left-wing | right-wing) bias, things will be.. difficult. 

Now, a discussion should be had about how the ML models are trained and if there are any biases in the labeled datasets that are used to drive recommendations, detect abusive content, etc. This is a complex problem, however! 

An important effect of the pervasive deployment of ML technologies is that it makes computing *probabilistic* instead of *deterministic*. i.e., we know what is likely to happen, but it is difficult to predict what *will* happen.

This paradigm shift makes it very difficult to point the finger at one or more woke/radical/reactionary programmer who decides to censor or advocate for free speech. 

Mr. Musk knows all this, of course. The entire Tesla “full self-driving” stack is built on ML. So, perhaps, a little bit of intellectual honesty might lead to a more interesting discourse about bias.

Links:
Why Elon Musk Wants to Open Source Twitter

Elon Musk’s Poll on whether the Twitter “algorithm” should be open-sourced: https://twitter.com/elonmusk/status/1507041396242407424

Twitter Engineering Blog: https://blog.twitter.com/engineering/en_us

MIT Technology Review has a good writeup about this: https://www.technologyreview.com/2022/04/27/1051472/the-problems-with-elon-musks-plan-to-open-source-the-twitter-algorithm/

Fiction: The Facility

First there was darkness and silence. Then the world turned a dim pink, and Kamal heard his heart beat, fast and irregular. The brightness grew until he opened his eyes to find a cluster of bright white lights shining down on him. He was lying on his back, and appeared to be alone in a room with whitewashed walls shrouded in gloom. He lay there on blinking at the bright lights, feeling like he had fallen off the top of a tall mountain and had landed having broken all the bones in his body.

He felt pins and needles all over his body. He tried to flex his fingers and twiddle his toes but nothing seemed to happen. He was alone and the room was quiet. He could hear a low hum in the background, but it was hardly noticeable above his labored breath. He could not recall how long he had been in that room. He had been drifting in and out of sleep.

Kamal dreamt of being the centre of attention in a large party. People congratulating him, shaking his hands and some hugging him. But it all felt very vague, as if something that had happened far enough to someone he barely knew. He felt some movement in the room, a slight shift in the air. He opened his eyes, and saw a shape, about the size of a man, standing in the gloom outside the cone of light from the lamp overhead.

He tried to speak, to call out to the shape, but he couldn’t. His throat felt raw, his lips dry and stuck together. He tried waving his arms, but the wave didn’t progress much more than a twitch of his fingers. More movement, and then his bed began to tilt up. It creaked slowly, raising Kamal to a sitting position.

The figure in the gloom moved forward. Kamal heard the familiar whine of electrical motors and the click of mechanical limbs. A bipedal drone strode into view and stopped a meter away from Kamal’s bed. The drone looked similar in design to the delivery drones that he had seen come and go from his father’s store back in London.

The drone held a brown box and a tray with a tumbler filled with water. It moved forward, and left the items on a table next to Kamal’s bed before disappearing back in the gloom. When he saw the water, he realised just how thirsty he was. It took him a few minutes to figure out how his limbs worked, but he managed to reach over and get himself a drink.

He had a look at the box the drone had left. There was a note attached to it, it was printed:

“You have just finished a long journey. Take some time to rest and recuperate. Keep hydrated. You will hear from us soon.”

Kamal opened the box and saw what looked like granola bars. He managed to eat one before falling back asleep.

This cycle continued a while. The drone would show up with water and food; always granola bars, sometimes apples and oranges.  Kamal’s time was spent eating, drinking and sleeping. He could barely remember his name.

After about 24 hours of waking up, the Kamal felt well enough to leave the bed and to explore his room. It was about five meters long and four across. There was a door leading to an empty corridor, and another door leading to a bathroom. The corridor was dark. The only light coming from the lamp above Kamal’s bed and from the small lamp in the bathroom.

Kamal saw himself in the mirror. He looked washed out, shrunken, awful. There were needle marks on his arms and his body simply refused to cooperate in doing the most simple tasks. Walking around, using toilet, all of it was an ordeal. Even worse was the confusion. He knew his name was Kamal, he came from London, his father owned a store, and lived with his brother in London. But he couldn’t tie his memories to his current situation.

The drone was his only companion, but like most drones, it was not particularly companionable. Kamal tried to talk to it, ask it where he was, what was going on, and got nothing. He tried blocking it from leaving, but it would just stand mute and still until Kamal got out of its way.

Kamal decided to follow the drone out of the door and into the corridor. The drone led him through the gloom to a circular security door. Kamal had seen doors similar to these in the entrances to his office block in Uxbridge. The drone walked through, but the door would not let Kamal through. He stood there a little while, tried knocking on the door, even tried shouting to see if there was someone around. But there was nothing but silence.

The next time the drone showed up, it was carrying something that looked like an ancient tablet. It left it on the bedside table along with Kamal’s food and was gone, motors whining and the ambulatory mechanism clicking down the corridor. Kamal picked up the tablet.  The glass screen was dark. There was a single button mounted on the bezel. When he pressed button, the tablet came to life:

Hello. This is a non-interactive device. This device will sound an alarm in 8 hours. Please rest until then. Follow all subsequent instructions.”

Kamal tried prodding the button again, he tried swiping up and down, right and left and nothing happened. The tablet displayed the same text, and a timer counting down the eight hours. There wasn’t anything else to do but wait.

The tablet started buzzing at the appointed time. Kamal was ready. He had been spent the last hour sitting on the bed watching the clock on the face of the table count down to zero. At the appointed time, the tablet lit up:

“Please walk down the corridor and through the security door. Please follow the next corridor to the end. The door at the end will be unlocked. Please enter the room and await further instructions.”

Kamal had been in the room for around 48 hours. He could just about walk, and he felt his mind clearing. The confusion receded a little. It had been replaced by agitation. This place was wrong, Kamal wasn’t supposed to be here. He was, however, glad to be given a chance to leave.

He walked out into the dark corridor and felt his way to the security door. It let him through and Kamal found himself in a well lit room. The room was dominated by a large screen set on the wall on one end. It was showing some text:

“Hello. This is an interactive screen and responds to voice commands.”

“Who are you? Why am I here? Where am I?”

“I am an interactive screen and respond to voice commands. I am programmed with a limited number of responses to your questions. Please speak slowly and clearly.”

Kamal had dealt with customer service bots before. The key was to be clear and unambiguous. “Where am I?”

“You are in a medical facility.”

“Why am I here?”

“You are recuperating from a medical procedure.”

What medical procedure?”

You were in stasis. You have been revived.”

Kamal’s confusion returned. Stasis? That was the procedure they had used for the terminally ill. Kamal felt almost a physical jolt as some of his memories came tumbling back. The party, the journey – the upload lottery had come through. He had heard that the transition could be difficult. The laws didn’t allow for two way communication between the Uploaded and the rest. Perhaps this was normal?

“The upload procedure… did it work? Where am I?”

“I am unable to answer that question. You are in a medical facility.”

Kamal felt a sense of dread rising. He had signed up for the procedure as soon as the lottery had come through. Perhaps he was dreaming.

“What do I do next?” Kamal asked.

“You are free to leave. Please keep the tablet with you at all time. Await further instructions.”

Leave what?”

“This facility.”

Kamal knew that the conversation wasn’t going to go anywhere.

“Where do I go?”

“Please leave this through the door on your right. The exit to the facility is via an escalator at the end of the next corridor.”

Is there someone I can talk to here?”

“I am programmed with a limited number of responses to your questions. Please speak slowly and clearly.”

Kamal knew he had to leave the facility if there was any hope to getting answers. He looked around the room and saw the shelves lining the walls were stacked with equipment. He saw boxes of the granola bars, some clothes and a sturdy pair of boots that just about fit him. He was still in a hospital gown. He would need to change and gather some supplies before leaving.

After about thirty minutes, Kamal took the elevator. He had found a small pack that he had loaded with granola bars and apples. The elevator creaked upwards before stopping at the only floor marked on the dashboard. Kamal left and walked out into bright sunlight.

The elevator had taken him up to an atrium. Or more accurately, what might have been an atrium. The doors opened on to a wide hall that had been taken over by vines. The sunlight was streaming down through the ruins of a skylight in the ceiling of the atrium. The doors, once made of glass and steel, were shattered and flung open on to a meadow. The atrium looked abandoned and it was quiet.

Kamal checked the tablet to see if something had changed. It remained dark. Kamal walked towards the doors, picking his way through the plants and shrubs that had colonised the concrete of the atrium floor. The sunlight felt good after the hours spent underground.

Kamal made his way out towards the meadow when he saw the girl. She looked like a teenager, perhaps fifteen or sixteen. She had dark brown skin and curly hair cropped close. She wore a loose shirt, a pair of denim trousers and sturdy boots. She stood facing him, mouth open in surprise. Clearly she hadn’t been expecting to see Kamal walk out of that ruined atrium.

“Hello. My name is Kamal. What is your name?”

The girl seemed to gather her thoughts. She took a step backwards but kept her eyes on Kamal.

“Do you speak English?”, it was a ridiculous question and Kamal knew it. But he couldn’t think of anything else to say.

“Yes”, she replied. “My name is Lily.”

 

Review: The Martian by Andy Weir

The MartianThe Martian by Andy Weir
My rating: 4 of 5 stars

I have a man crush on Mark Watney. There, I said it. If you think this makes my review biased, move on sir, move on.

Still here? Good. Let me tell you about my mate Mark. He is the nerd’s nerd. Not content with just being a kick-ass engineer / astronaut, he can farm, he can drive, heck he can even sew.. on Mars! He possesses a testicular fortitude and self belief not often seen in these post modern, post apocalyptic morass that calls itself modern science fiction. What a dude!

I won’t lie to you, I was dubious about this book. After much prevarication, I finally booked a ticket to Mars on the enthusiastic recommendation of a normally rather taciturn colleague.

So what is The Martian all about? Its about an astronaut stuck on Mars after a NASA mission goes horribly wrong. We follow Mark Watney as he figures out how to survive on a desolate, inhabitable and bleak planet. Most of the book is written as a series of log entries and monologues from Mark as he rolls with the punches Mars throws at him and comes up with ingenious solutions to problems that would have stumped us mere mortals. He manages all of this while listening to disco – yes, disco. Mark has it rough.

There is a lot of science and a lot of engineering in this book. We get long passages on how to extract water from rocket fuel, how to grow potatoes in space and even how to build a crude sextant to navigate on a planet that does not have a magnetic field. The book could have been tremendously boring but the main character is likeable and the writing in places is wonderfully, crudely funny.

There are passages that are set on Earth and in space that provide a bit of context around the plot. Honestly, I skimmed through some of these chapters to get back to the action on Mars.

So yes, I am a fan of The Martian. Yet, it only gets 4 / 5 stars. Why? Well, you never get a real sense of danger for Mark. Yes, the situations are difficult and bad things happen. Yet, you just know that Mark will survive. This drains some of the suspense out of the book. But hey, its a minor quibble!

If you are a fan of science fiction, you will enjoy The Martian. If you are a space nerd and you Buzz Aldrin is your hero, you will love it. If you still happily watch re-runs of McGuyver, go and buy this book now!

View all my reviews

The WhatsApp acquisition

The water-cooler was abuzz this morning with news of Facebook’s $19 billion acquisition of WhatsApp, a tiny company. With the claimed 400 million users that WhatsApp brings to Facebook, the numbers involve value each user at $40. That is an astonishing amount of money for a service that is monetised through application sales, not via advertisement. There have been a number of articles and blog posts online analysing this deal. This is not one of them..

My colleagues are a quiet and taciturn lot. Office banter is limited to a “Good Morning” and a “See you later..” outside of the lunch hour. For the first time, in my admittedly short stint here, we had a bonafide conversation that was not even tangentially related to trading systems and market data feeds. We got talking about what it means to be a programmer working outside of the startup / silicon valley scene. One of my colleagues remarked that he spent half a decade in further education and a lot longer learning the ropes until he got to the point now where he is comfortable and financially secure. He wondered if that time would have been better spent writing a new chat or social network. Perhaps a new way of optimising the transmission and sharing of ribald jokes, or for improving the sexting workflow.

We carried on in a similar vein for a while when the most introverted of our lot spoke up. He said: “I was just never interested. The thought of building the next Facebook or Twitter just doesn’t excite me. It was never something that was on my radar.”

I spend way too much time on Hacker News. The Silicon Valley culture and eco-system fascinates me, but it does not inspire me. I marvel at the numbers that are thrown around. A few billion here, a few billion there, but I also wonder about the utility of it all. It is now fashionable to talk about how much of a talent drain banking has become. How so many people left promising careers in academia and engineering to cut code and make money on Wall Street and the City. In a few years I can see people talking in similar terms about Silicon Valley. “He was a promising scientist, but he joined Google to help them optimise the placement of adverts on search results.”

I find the earnest tone of discussions on Hacker News and of the job postings for these start ups deeply ironic. They talk about changing the world, wanting rockstars and working on cool new technologies. Yet, the end goal is a big payout via IPO or acquisition having built a better way of sharing food selfies. I think these headline acquisitions are a honey trap for programmers. Somebody, like my colleague, who wouldn’t really even think about working for a startup building a “trivial” app might realise that the App may be a gateway to that long dreamt of retirement on the beach.. You might get a lot more people ready to work for peanuts with the hope of striking it rich one day. Perhaps it is not a colossal waste of money after all..