Uncle Sam as Entrepreneur

Photo by Min An on Pexels.com

“Before Covid-19, the record for the fastest vaccine development — for mumps — was four years. Most vaccines have required more than a decade of research and experimentation.

Yet yesterday morning, less than a year after the discovery of Covid, a critical care nurse in Queens named Sandra Lindsay became the first American to participate in the mass vaccination program for the coronavirus. ‘I feel like healing is coming,’ she said afterward.

It is a stunning story of scientific success.

It also fits a pattern that stretches back decades: Many of the biggest technological breakthroughs in American history have not sprung from the private sector. They have instead been the result of collaboration between private companies and the federal government.

The Defense Department, after all, built the internet. Government research and development also led to transistors, silicon chips, radar, jet airplanes, satellites, artificial limbs, cortisone, flat screens and much more, as the M.I.T. economists Jonathan Gruber and Simon Johnson point out in their recent book, “Jump-Starting America.”

‘Almost everything about your computer today — and the way you use it — stems from government funding at the early stages,’ Gruber and Johnson write.

Why? Because basic research is usually too uncertain and expensive for any one company to afford. Often, it isn’t even clear which future products the research may create. No kitchen appliance company ever would have thought to do the military research that led to the microwave oven.

With Covid, the vaccines from both Pfizer and Moderna rely on years of government-funded (and sometimes government-conducted) research into viral proteins and genetics. That research, Kaiser Health News explains, is ‘the essential ingredient in the rapid development of vaccines in response to Covid-19.’

The federal help accelerated this year. The government funded Moderna’s work in recent months, as part of the billions of dollars it spent to make possible a record-breaking vaccine, The Atlantic’s Ed Yong writes. And while Pfizer turned down direct federal funding, it asked for the government’s help in procuring supplies and also signed a $1.95 billion ‘advance purchase’ agreement with Washington.

As my colleague Neil Irwin has written: ‘The nine months of the pandemic have shown that in a modern state, capitalism can save the day — but only when the government exercises its power to guide the economy and act as the ultimate absorber of risk. The lesson of Covid capitalism is that big business needs big government, and vice versa.’

What are the lessons for the post-Covid world? Solving the biggest challenges, like climate change, will almost certainly depend on a combination of public-sector funding and private-sector ingenuity.

Yet as Gruber and Johnson note, federal funding of science has become a smaller part of the U.S. economy than it used to be. Which means the Covid vaccine is both an inspiring success and something of an exception. ‘On its current course,’ the economists write, ‘America seems unlikely to continue its dominance of invention.’

— The New York Times Morning Briefing

New Austrian ‘e-prop’ Learning Algorithm Could Expand AI Applications

Photo by Pixabay on Pexels.com

“The high energy consumption of artificial neural networks’ learning activities is one of the biggest hurdles for the broad use of Artificial Intelligence (AI), especially in mobile applications. One approach to solving this problem can be gleaned from knowledge about the human brain. Although it has the computing power of a supercomputer, it only needs 20 watts, which is only a millionth of the energy of a supercomputer. One of the reasons for this is the efficient transfer of information between neurons in the brain. Neurons send short electrical impulses (spikes) to other neurons – but, to save energy, only as often as absolutely necessary …

With many of the machine learning techniques currently in use, all network activities are stored centrally and offline in order to trace every few steps how the connections were used during the calculations. However, this requires a constant data transfer between the memory and the processors – one of the main reasons for the excessive energy consumption of current AI implementations. e-prop, on the other hand, works completely online and does not require separate memory even in real operation – thus making learning much more energy efficient.”

— New learning algorithm should significantly expand the possible applications of AI, Human Brain Project