AI: Productivity vs Performance

Exploring a world when more work will be done by freelancers, solopreneurs and outcome-based hiring

Do we think of these in the right order?

We are living in a world where technology is a double edged sword.

On one side, it democratizes access to information, education, connection and increases competition.

On the other side, it pushes businesses and individuals even further in a ‘performance’ and growth at all costs context.

The future of work in the age of AI will be extremely polarized: offering both unparalleled opportunities and exacerbating existing inequalities.

---

The rise of disparity

Let’s start with the famous ‘Elephant’ chart of economic growth (also featured in Andy’s great article ‘Five Graphs that Challenge the Way we Work’).

Why an elephant, you say? It has a hump that includes the world’s poorest, mostly made up of people from developing countries; a valley that contains the working and middle class of the developed world, but also the upper class of poorer countries; and a “trunk” made up of the global elite.

The hump and subsequent valley on the left illustrate the enormous growth emerging countries saw relative to the middle and working classes in developed countries. This difference in growth has driven global inequality down to levels not seen since the 1700s.

But the new estimates also revealed that the richest 1 percent saw more growth than any other income level—resulting in the elongated trunk on the right of the chart.

How the elephant’s trunk will move is the key element we need to watch out for, especially in the light of AI.

Productivity vs Compensation

When we think of AI, we think of productivity. But who benefits from productivity gains?

The answer can be found in this graph, showing how hourly compensation has been closely falling the productivity growth up until 1980, and then started to decouple. So supposedly, if that productivity actually created wealth/value, it ended up in the 1%.

Now there’s several ways to look at this data. ‘The wages have grown!’ some will say. True, but cost of living has risen dramatically too. This means, that adjusted to inflation, the widening gap has not played out in favor of workers, but more to businesses.

Just think of how technology impacted internet economics; more than 60% of ad spend outside of China goes to Amazon, Meta and Alphabet. If AI is dominated by the same companies (plus OpenAI), we’re still going to be putting the majority of gains in the hands of the few.

What about productivity at work, you say?

In this transition, being more productive with AI does not imply growth of compensation. From my point of view, the only way that this could really work, is if organizations turned to a more ‘Web3-like’ distribution of profits, or if work was paid by outcome.

Right now, all businesses operate and compensate based on time. Time-based pay was straightforward to calculate and administer during the industrial revolution, making it suitable for businesses that needed to employ staff for general roles where productivity was hard to measure.

Time-based salaries are going to be out of date for several reasons:

1. Lack of Incentive for Productivity: Time-based pay does not encourage greater productivity since employees are paid regardless of their output level. This can lead to inefficiencies if some workers produce more value than others without additional compensation.

2. Modern Work Environments: With advancements in technology and changes in work environments (e.g., remote work), measuring productivity by time alone becomes less relevant. Output-based salaries can better align with modern performance metrics that focus on results rather than hours worked.

3. Global Competition and Innovation: In today's competitive global economy, companies often seek ways to motivate employees beyond just paying them by the hour or month. Performance bonuses or output-based incentives can drive innovation and efficiency more effectively than traditional time-based models.

What happens with an ‘output’ based compensation?

Let’s say we’re living in 2030 (or 2035) when AI has already automated most of the white-collar tasks it could take. We can imagine one scenario where history simply repeats itself (new jobs are created, but we still employ and compensate in the same way as today) or an alternative scenario, where 50% of the compensation is time-based and 50% is outcome based.

This could be broken in three main components:

1. Outcome-Based Metrics:

   • Project Completion Rates: Focus on how efficiently tasks are completed rather than the time taken

   • Milestone Achievements: Track progress by identifying key milestones within projects

   • Quality of Deliverables: Assess the quality and punctuality of project outcomes

2. Feedback and Peer Assessment:

   1. Use 360-degree feedback to evaluate productivity based on peer comments and interactions.

3. Token Economy - Many Web3 companies use tokens (like digital ownership shares) as part of their compensation packages. This allows employees to benefit directly from the project's growth, aligning their interests with the company's success (more on this in ‘The Decentralised Workforce’ and ‘Work3: The Internet of Careers’)

Clearly, there’s also downsides to this solution like any, but any company that would bake these elements into their operating and people model could harness the power of smaller teams, but with higher level of skin in the game and distributed / fairer incentive.

Last, but not least, I imagine a future where:

Freelancers are the majority of workers - Statista (chart above) estimates that in 2028 just in the US they will be 90 million, 51% of the total workforce.

Solopreneurship will be taught in schools and half of workers will have tried to launch their own business.

Tools and Businesses will be paid for results - The more digital services and skills become commoditized, the more SaaS tools and service businesses will need to use AI productivity to drive new business models and compete for clients.

Case studies: Outcome-based pricing in action

Let’s examine some real-world examples of companies successfully implementing outcome-based pricing models.

Riskified, an ecommerce fraud prevention provider, uses advanced algorithms to assess transaction risks in real time, providing instant approve/decline decisions. Riskified guarantees approved transactions against fraud and charges only for successfully approved, fraud-free transactions. This model incentivizes Riskified to continually improve its algorithms and deliver value, as the company only profits when its solution effectively prevents fraud and increases sales.

Riskified’s Chargeback Guarantee is prominently featured on its website

Source: Riskified

Riskified’s approach is particularly effective because it directly reduces client expenses. This makes outcome-based pricing more appealing, as customers can easily quantify savings and see a link between the service provided and its financial impact.

iDenfy, another player in the fraud management space, offers KYC (Know Your Customer) verification services with a similar outcome-based pricing model. It helps businesses verify customer identities and prevent fraud, charging only for successfully approved users. Both Riskified and iDenfy showcase how fraud prevention and identity verification services are well suited for outcome-based pricing, as success is clearly defined and measurable, directly linking service to value.

Shifting from fraud prevention to customer service, Intercom’s AI chatbot, Fin, offers another innovative example of outcome-based pricing. Launched in 2023 and closely watched as an early AI-based implementation, Fin costs $0.99 per successful resolution. This model, where clients pay only for effective AI interactions, demonstrates how companies can blend traditional usage-based and outcome-based pricing.

Intercom’s July 2024 pricing for its AI agent, Fin

Source: Intercom

As an early adopter in AI services, Intercom and its approach may accelerate a broader trend. AI’s ability to process vast data and deliver measurable results makes it particularly suited to outcome-based pricing, potentially reshaping pricing strategies across various SaaS sectors.

Productivity vs Human Connection

One last consideration about time and work. Until we have people that compare humans with machines, and think that human productivity is a simple multiplier of time, we’re doomed as a society. Think of it: the man behind the mission of AGI and automation of everything, relies on 120 hour work weeks from humans. There is scientific evidence suggesting that beyond a certain threshold (typically around 40-50 hours per week), additional hours worked do not necessarily translate into increased output but rather lead to decreased efficiency per hour due to fatigue. There is also scientific evidence that multitasking can decrease productivity by up to 40% due to the time spent switching between tasks. This switching process disrupts information processing, leading to decreased memory and higher error rates.

All of this can lead to mental fatigue and even negatively affect decision-making skills by reducing focus and creativity: exactly the two vital skills we will need to thrive in the future.

Until we get that right, in culture, in media, in organizational models, we will just be hamsters running around a wheel, with a tip of machoism and populist rhetoric masking our stagnation.

Successful societies can respond to this by a) adapting b) putting humans at the center c) thinking of equitable access. That’s why lifelong learning, support for entrepreneurialism and new mechanisms for financial securities need to be designed at the core of culture, and public support - not just at the private level (all of these widely covered in our articles ‘Reinventing Work for the Digital Golden Age’ and ‘Building a Better World Without Jobs’.

Last order of the day, for fun purposes only (so feel free to stop here if you’re not in for the treat) - the response from Gemini on how the Elephant shaped trunk could change with the impact of AI:

• The Dromedary: A single-humped camel with a massive hump representing the ultra-rich, while the rest of its body is scrawny and malnourished, symbolizing the struggling masses.

• The Platypus: A bizarre creature with a duckbill, beaver tail, and otter feet, representing the confusing and unpredictable nature of the AI-driven economy, where traditional job categories blur and new, strange ones emerge.

• The Bloated Tick: A grotesque creature swollen with wealth, sucking the lifeblood out of the rest of the economy, symbolizing the parasitic nature of extreme wealth concentration in the age of AI.

• The Three-Headed Cerberus: A mythical beast guarding the gates of wealth, with each head representing a different sector of the AI-powered elite: tech giants, financial wizards, and political puppeteers.

• The Invisible Man: A ghostly figure representing the vast majority of the population, rendered invisible by the overwhelming power of AI and the elite, struggling to survive in a world where their skills and labor are no longer valued.

Tried to ask it for a sketch, but this is all I got.

Ciao,
Matteo