A new White House report presents a vision of research powered by artificial intelligence agents and robotic labs. While promising a dramatic acceleration of science, the demand for political accountability raises concerns about government interference.
Eighty-one years ago, Van Nievera Bush wrote one of the most important documents in the history of science. In the document, Bush warned against the human tendency to support only scientific projects with clear and immediate applications, and proposed instead the establishment of the scientific community as we know it today. Thanks to that document, the National Science Foundation was established, which supported basic scientific research.
And what an important decision this was.
The scientific research the foundation has supported since then, which seemed shoddy and useless at the time, has borne enormous fruit. Today, you surf the Internet, built on algorithms and protocols developed with foundation funding, using smartphones that contain semiconductors courtesy of foundation funding, and operate artificial intelligences based on artificial neural networks that also received foundation funding.
And it all started with a Bush document, which was aptly named - "Science: The Infinite Frontier."
Why do I mention such forgetfulness? Because at the end of July it was published A new document from the White House Office of Science and Technology Policy, entitled Science: A New Golden Age, which presents a new vision for American science. It is an exciting, moving, and bold document—and more than a little depressing. Whether you agree with the future it describes or oppose it, one thing is clear: it cannot be ignored. This document will make history.
But what history exactly?
Let's start with the positive.
Labs of the Future
“Imagine an investor posting a $1 million reward for the first validated medical target for a rare disease,” the paper’s authors describe the future they envision for us. “An [artificial intelligence] agent investigating related problems spots a promising clue and immediately posts a smaller reward to be given to whoever confirms its findings. Other agents assess whether they can tackle the problem, compete for the research, and hire an autonomous lab over the network. The lab automatically runs the experiment and returns signed, encrypted results.”
And what happens then?
“The [original] agent examines the evidence, updates its models, and publishes the conclusions in a shared research journal. […] Smart contracts automatically release budgets when the results are validated.”
In the future described in the document, the vast majority of research in laboratories is carried out by agents controlling robots. There are agents who specialize in reproducing experiments to ensure that the original results are correct. Other agents focus on tiny incremental discoveries, with small, measured progress with each experiment, while their peers concentrate on big breakthroughs—with a high risk of failure.
In this world, “cloud labs are becoming the factories of the economy.” Already today, robotic facilities exist that can synthesize molecules, run tests, and return results without human intervention. The authors predict that these facilities will become network hubs that any agent can operate for the appropriate fee.
"An AI testing a hypothesis about protein folding could rent a lab in Colorado, run crystallography experiments, get results within hours, and feed them into its next round of thinking, while collaborating with humans in Boston."
And the end result? Physical experiments, which constitute the main bottleneck of experimental science, will be just as accessible as computing power is today. And science will leap forward.
That's all. Nothing big.
But how do we do all this?
The big change
To bring us to this promising future, the authors of the document propose several reforms to science in the United States.
They believe that collaborations should be encouraged, for example between artificial intelligence researchers and researchers in all other fields, to allow artificial intelligence to advance any topic – and perhaps also acquire the knowledge it needs to conduct research itself.
“Ultimately,” they write, “we will need scientific institutions with AI embedded within them, and with procedures, infrastructure, funding models, and cultural norms built around the use of powerful AI systems.”
All this AI activity requires a lot of energy and a lot of computing power. All of this will come from the private sector, which will collaborate with national laboratories to produce research that will serve both the public and private companies. Thanks to this research, we will find new ways to produce energy, such as nuclear fusion, and of course, to improve AI.
And all of this is starting to happen today, as part of the “Genesis Mission,” which Trump announced in November 2025. The goal of “Genesis” is to build the American “Unified Platform for Science and Security,” which will connect supercomputers, artificial intelligence systems, scientific tools, and databases. All of these will become “a single discovery engine that will double the productivity and impact of American science and engineering within a decade.”
To meet this overarching goal, the scientific establishment must understand that there is no longer room for separate, isolated laboratories. Every scientific tool embedded in a laboratory must be redesigned to adapt it to full automation, with open interfaces and standards that are adapted to industry standards. And – “Beresheet” is currently investing in 14 projects designed to move laboratories to full automation, with autonomous remote control of the experiments being conducted in them.
One problem that could arise with all these impressive tools is that they will work too well. That is, they will conduct a huge number of experiments and produce a similar number of results. The scientific literature will be flooded with results, not all of which will necessarily be correct. In other words, it will be flooded with garbage.
To prevent this future, one of the important conditions of “Beresheet” is to promote tools that will allow experiments to be easily and efficiently reproduced. Along with this, also the full sharing of the information, data, and methodologies – the research methods – behind each experiment. In this way, any new experiment can be automatically reproduced at low cost by autonomous laboratories throughout the United States.
And all under the direction and supervision of the United States government.
A government to integrate them and their interests, a government to attract them and their interests
If the document had ended with only the steps I described, I would have gone out to celebrate its publication in the streets and invited all of Israel to a beer at my expense. The authors describe a wealth of real failures in the scientific establishment in the United States and in general, and propose serious steps to address them. And if that were not enough, they dare to foresee a revolutionary future for science and human society.
As one of the cultural figures in Israel says: applause. They deserve it.
But the document does not begin or end there.
The most disturbing part of the document comes right at the beginning, when the authors state that science should be subordinate to politics. Or as they put it:
"Federal support for science should be politically accountable."
What does it mean, "subject to politics"?
Great question. And the authors don't give a good answer to it. They make it clear that the intention is not for politicians to dictate how research should be conducted, which is excellent. They do say that science today is subject to many different factors and interests, and there is certainly some truth to that. Pharmaceutical companies, for example, fund medical research; universities with a social agenda promote progressive-liberal or conservative-religious research. It is clear that cases in which these factors are too involved in the scientific process and negatively influence it should be condemned.
But is the solution really to bring politicians into this cauldron, so that they too can influence the directions of research? Should we now rely on politicians to be balanced and honest and to strive for scientific truth and nothing more? Or is it perhaps more likely that politicians will just add another layer of corruption, nepotism, and the subordination of science to temporary political needs?
If I had to bet on the experience of other countries where politics mixed with science—such as the Soviet Union and communist China—the latter outcome is much more likely. Wherever the government allowed itself to exert too much influence over science, certain scientists became favorites of the government, thus gaining enormous power over their colleagues. In the case of the Soviet Union, the result was catastrophic: the agronomist Trofim Lysenko won the support of the government, rose to the top of the scientific pyramid, spread unfounded “scientific” theories, and persecuted those who opposed him. Dissidents were dismissed, imprisoned, and in some cases even died in prison.
Want to guess what happened? The Soviet Union adopted Lysenko's "science" and implemented harmful agricultural methods, which damaged crops and food security. Similar ideas were also adopted in China during the "Great Leap Forward", as part of a broader policy that ended in mass starvation and the deaths of millions.
Is this also what will happen in the United States? We don't have much choice but to wait, see, and hope. To hope that the government will finally understand the limitations of power, and let science advance the country and humanity.
And if we succeed, we will win the biggest prize.
The future promised to us
On the last page of the report, the authors dare to predict the best future that science can promise us.
“The 21st century will be built on materials engineered from first principles, on metamaterials that bend light in ways that nature has never attempted, on matter that can be programmed on command, and on structures that assemble and grow like living things but serve engineering purposes,” they write. “Perhaps we will begin to engineer cells precisely as we engineer electrical circuits today, and program immune systems to hunt down cancerous tumors […] and shape tissue growth to repair damaged organs.”
"If we do all this right, within a generation the diseases that currently kill millions – such as heart failure, neurodegenerative diseases and cancer – could succumb one by one to the tools we are now starting to build."
But will it happen? That's up to us.
"This is the task of our generation," the report's authors explain: "to design the institutions, standards, and capabilities required to direct a scientific establishment that is larger, faster, and less comprehensible to a single individual than any other in history. And when we do so, we will determine not only the future of America's technological capabilities, but also the trajectory of human knowledge itself."
Questions and Answers
What is the report "Science: A New Golden Age"?
This is a report by the White House Office of Science and Technology Policy, published in July 2026, that presents a vision for transforming the United States' research and development system. It deals with, among other things, artificial intelligence, autonomous laboratories, research funding, and the relationship between science and government.
What is the "Genesis Mission"?
The Genesis Mission is an American initiative designed to connect supercomputers, artificial intelligence systems, scientific instruments, and databases into a common research platform. Its stated goal is to double the productivity and impact of American science and engineering within a decade.
How are autonomous laboratories supposed to operate?
AI agents will be able to propose hypotheses, order experiments in networked robotic laboratories, analyze the results, and plan the next experiment. Other laboratories will be able to automatically replicate the experiments to test their reliability.
Why does the new vision also arouse concern?
Along with promising to accelerate research, the report states that federal support for science should be “subject to politics.” This wording raises concerns that politicians and governments will influence the choice of research areas and the distribution of budgets according to temporary interests.
More on the subject on the science website
- Self-researching artificial intelligence
- The end of the era of human research? Artificial intelligence is researching itself and developing its own next generation
- Trump's grant rules could isolate American science
- 250 years of the United States: This is how the world's greatest scientific power was built — and how it endangers its future
- What does Trump want from universities?
For the original publication: Opening the original publication
