Einstein, Dr. House & Harvey Specter — At your beck and call

We’ve Seen This Movie Before

There was a lot of hype when Netflix launched its streaming service in 2007, but the reality came up short: endless buffering, limited content, and low resolution made streaming a poor stand-in for DVDs. Industry executives scoffed, with Blockbuster's CEO Jim Keyes saying he was “frankly confused by this fascination everybody has with Netflix” and doubling down on physical media.

Yet Netflix’s CEO, Reed Hastings, saw beyond these glitches. Rather than judging streaming’s potential by its present state, he looked at bandwidth’s trajectory, concluding that connectivity would soon become functionally free and infinite. He bet his company on that glide path, and we know how that turned out.

Coming to a screen near you: Free super intelligence.

Blockbuster’s mistake sprang from two stubborn biases: Exponential-Growth Bias (we underestimate compounding curves) and Status-Quo Bias (we assume tomorrow looks like today). In the mostly static world we evolved in, these instincts served us well. In the rapidly changing world we live in, they lead us astray.

Too often people size up artificial intelligence by its hallucinations and limitations, with our minds wired to think that since GPT-4 cannot truly replace lawyers, doctors or coders today (which it cannot), all is well. But assessing AI based on a random snapshot in time (today!) is a blockbuster blunder. Instead, trace AI’s progress over recent years, then extend the curve a few years out, and you’ll see that intelligence is on a path to become free and infinite.

In just the past 18 months, the cost of running frontier models has collapsed by more than 95 percent (cost-per-million-tokens), even as their skills vaulted from high‑school essays to post‑doctoral reasoning. Fast forward that movie, and you glimpse a world where elite cognitive ability is as cheap and ubiquitous as Wi‑Fi. That changes everything.

“Where We’re Going We Don’t Need Roads”

Labor markets, education systems, tax codes—virtually every institution—presume that expertise is scarce and slow to produce. Attorneys spend a decade or more mastering their craft before arguing at the Supreme Court; doctors invest still more time before patients trust them with complex cases; even an average software developer earns multiples of the median wage because experienced coders are hard to mint. But if intelligence becomes free and infinite, the educational, corporate, and fiscal scaffolding we take for granted buckles.

Four converging exponential curves are driving the price‑performance of AI toward zero and its reach toward infinity:

• More chips. AI companies are buying millions more chips each year, roughly doubling their global fleet annually.
• More powerful chips. Each new chip delivers about twice the raw compute every 18–24 months.
• Better algorithms. Smarter algorithms extract roughly 2× more performance from the same hardware every 9-12 months.
• Other optimizations. Model‑shrinking techniques and purpose‑built accelerators slash the cost of each answer by about half every few months.

All that doubling and halving compounds, so that the effective “muscle” of state‑of‑the‑art AI increases 5-to-6-fold annually, about an order of magnitude faster than the breakneck pace of Moore’s law. That hurtles us towards a future where synthetic brains outperform biological ones at virtually every task.

Daniel Schreiber.

“I Know Kung Fu”

Machines won’t just think smarter, faster and cheaper than humans, they will ingest, share and pool their knowledge in alien ways. Human bandwidth for knowledge transfer—speaking, writing, reading—tops out at two dozen bits per second. Digital knowledge diffuses at billions of bits per second.

When Tesla’s autonomous driving system masters a novel construction detour, that skill is pushed to eight million Teslas in no time. Training that many human drivers takes about 100,000 years in aggregate. Biology drip‑feeds where software blasts, and that gulf is set to erase the scarcity of cognition.

We will all feel the quake. Wage premia in medicine, software and finance will compress, and payroll taxes will sag. Location will matter less, as expertise moves to the cloud; and corporate size will no longer guarantee dominance, as nimble newcomers with rented cognition challenge incumbents once shielded by head‑count. Permanent unemployment may surge.

On the flip side, once synthetic brains outperform the biological kind, the world's very best doctors, psychologists, tutors, and entertainers will be on tap to everyone, everywhere, anytime and for free–or close enough. Think of it as having the likes of Albert Einstein, Gregory House, and Harvey Specter at your beck and call 24/7. Once AI does the work of lawyers, accountants, managers, marketeers and drivers for next to nothing, the costs of everything will come down, and–if we navigate this well–abundance will become… abundant.

It is natural to recoil when faced with changes of this magnitude, but that’s our status-quo-bias talking. This is no time for heads in sand. Like General Shinseki said, “If you don’t like change, you’ll like irrelevance even less.”

Israel is small enough to pivot quickly and ambitious enough to aim high. We need to start acting along two complementary tracks: growing the pie, and ensuring everyone gets a slice.

“Show Me The Money”

Growing the pie means ensuring Israel captures as much of the AI abundance as we can, and here the Startup Nation has some catching up to do. The following 4 steps may serve as a helpful roadmap:

Regulate performance, not people. Israel, historically, has simply not been an attractive jurisdiction in which to develop and pilot AI. It’s time to adopt regulations that pivot from credential‑based licensing toward performance‑based benchmarks. Simply put, when an AI system demonstrably outperforms humans at driving, lawyering, diagnosing, and counseling, regulation should allow it to chauffeur, advise, prescribe, and therapize.

Open public data—safely. A key differentiator in the age of AI will be access to proprietary, high quality data. Israel’s early digitization puts it in a privileged place in this regard, and it's time to draft privacy‑preserving standards to anonymize and release health, legal, and geospatial datasets, so Israeli models can train on data few foreign rivals can replicate.

Government Moonshot: The government should undertake a big, hairy, audacious goal, along the lines of ‘cutting government spending by 20% over 5 years while doubling the quality of service to its citizens’. Such a BHAG is unachievable without wholesale adoption of AI, creating a forcing function that will spur transformations in healthcare, transportation, education, local government and much besides. Beyond lowering an overextended government budget, and improving government services, such a moonshot will fuel countless startups who will race to develop the requisite AI tools.

AI Independence. Commit to a multi‑year roadmap for domestic AI clusters, ensuring local researchers and startups are never priced out of the new industrial feedstock and that the country can withstand geopolitical shocks and boycotts without ‘losing its mind’.

“Nobody Puts Baby In a Corner”

Moving Israel to the forefront of AI adoption is essential given that we operate in a global market, but it will create major dislocations that need addressing. Now is the time to start debating, piloting, and stress‑testing policies to ensure that in the era of free and infinite intelligence, all Israelis are better off. Here are three thought starters:

Shift tax to what remains onshore. For example, explore transitioning tax revenue from labor to consumption. A roadmap to a gradual VAT increase coupled with a phased reduction of income and corporate taxes would be a solid beginning.

Distribute the AI dividends broadly. Pair the eventual shift in taxes with a universal basic income—framed as a citizen dividend from automation—to keep the system progressive, and ensure all citizens are better off, at least financially, in the age of AI.

Reimagine life beyond re‑skilling. Commission pilots for a social contract in which thriving is decoupled from employment, enabling citizens to devote time to caregiving, learning, art, exploration, and community—domains where presence, not professional abilities, matters.

“To Infinity and Beyond”

Netflix triumphed because it redesigned its business before the bandwidth curve peaked. Israel now stands at a similar crossroads, but the stakes this time are immeasurably higher. The choices are clear: behave as though we are “frankly confused by this fascination” with AI, and double down on practices and modes of thinking built for cognitive scarcity; or pivot our economy and institutions for the coming age.

You can be sure that tomorrow's winners—individuals, organizations and countries—will be those that choose the latter.

Daniel Schreiber is the Co-Founder and CEO of Lemonade.