It's easy to find gloomy predictions for continued slow growth of the US economy. Thus, Lee Branstetter and Daniel Sichel caught my eye with their essay, "The Case for an American Productivity Revival,"  written as a "Policy Brief" for the Peterson Institute of International Economics (June 2017, PB17-26). Here how they start:

Labor productivity performance in the United States has been dismal for more than a decade. But productivity slowdowns—even lengthy ones—are nothing new in US economic history. This Policy Brief makes the case that the current slowdown will come to an end as a new productivity revival takes hold.  
Why the optimism? Official price indexes indicate that innovation in the technology sector has slowed to a crawl, but better data indicate rapid progress. Standard measures, focused on physical capital, suggest that business investment is weak, but broader measures of investment that incorporate intellectual and organizational capital report much more robust investment. New technological opportunities in healthcare, robotics, education, and the technology of invention itself provide additional reasons for optimism. This Policy Brief gauges the potential productivity impact of these developments. The evidence points to a likely revival of US labor productivity growth from the 0.5 percent average rate registered since 2010 to a pace of 2 percent or more. A productivity revival of this magnitude would provide a solid foundation for steady increases in wages ... 

The essay spells out details behind these claims.  Here are a few of the comments that caught my eye.

Past methods for adjusting for the improved quality of microprocessors may not be working well at capturing changes in the last decade or so. 

Before the mid-
2000s, the posted prices of MPUs tended to fall as newer
models were introduced. This price trajectory allowed a
standard methodology used for semiconductors in the
producer price index (matched-model indexes) to capture
quality change through the rapid price declines of older
models. Since the mid-2000s, posted prices of Intel MPUs
have tended to remain stable, even after the introduction
of newer, more powerful models. Reflecting these relatively
flat price profiles, a matched-model index will indicate little
change in quality-adjusted prices even if the quality of each
newly introduced model is much greater than its predecessor.
The new price measure Byrne, Oliner, and Sichel
developed (an hedonic index) more fully captures ongoing
quality change and reveals rapid price declines after this
quality change is taken into account.
This evidence on faster price declines indicates that
innovation and multifactor productivity growth in semiconductors—
the general-purpose technology behind much
of the digital revolution—has been far more rapid than official
indexes suggest.

Although conventional tangible business investment is down as a share of GDP, intangible investment is on the rise.

In this figure, the light blue line shows tangible business investment, and the well-known pattern of overall decline (as a share of GDP) since the 1970s. The dark blue line shows the official US government statistical measure of "intangible investment," which includes "software, scientific R&D, mineral exploration, and the development of entertainment products." The dashed red line shows a broader version of intangible investment that includes both the official measures and also "nonscientific product development, brand equity, training, and organizational capital." As the authors write (footnotes and citations omitted):
In fact, the overall investment share of both tangible and all intangible capital has been relatively stable since the late 1970s. This conclusion is not surprising in an economy in which the newest technical capabilities and products rely at least as much on intangible capital as on tangible capital. This feature surely characterizes leading companies such as Google, Amazon, Facebook, and Microsoft. Even industrial companies like GE are increasingly investing in big data, predictive analytics, and machine learning.

In short, Branstetter and Sichel believe that the official statistics are understating both current productivity gains as well as the investments that firms are making for the future.  They then note: "Four developments have the potential to contribute to faster productivity growth in the United States: improvements in the healthcare system, increasing use of robots, a revolution in e-learning, and globalization of invention." They further argue that these changes can be supported by some mostly well-known policies: for example,
  • "robust federal investment in basic science"
  • "immigrant scientists and entrepreneurs play a disproportionate role in driving the technological advances that power productivity growth in the United States"
  • "globalization of invention presupposes the continuation of an open global trading and investment system supported by the United States"
  • "a public agency or public-private partnership that could certify the efficacy of new educational technologies in the same way the Food and Drug Administration (FDA) certifies the safety and efficacy of new drugs, by supervising rigorous, randomized control trials. Modest policy effort
  • in this direction could yield rich dividends in the form of much faster, more cost-effective human capital formation."
I would add that economic growth is by its nature a disruptive process, and part of embracing this disruption is to find ways for both its benefits and costs to be widely shared. The authors conclude: "A standard productivity growth accounting framework captures these factors to highlight how a significant revival of productivity growth could emerge, especially in the medium to long run. A pace of 2¼ percent a year is eminently plausible—and there are solid reasons to hope for
even more rapid productivity growth."