Minority Report Meets Drug Discovery: Intelligent Gestural Interfaces and the Future of Medicine

In 2002, Steven Spielberg’s Minority Report introduced one of the most iconic visions of the future: a world where data is accessed, manipulated, and visualized through an immersive, gestural interface. The scene where Tom Cruise’s character, Police Chief John Anderton, swiftly navigates vast amounts of visual information by simply swiping his hands through thin air is not just aesthetically captivating but also hints at the profound potential of such interfaces in real-world applications—particularly in fields as complex as drug discovery. Just like detective work involves combining and coordinating data from disparate sources such as GPS, real-time tracking, historical case studies, image recognition and witness reports, drug discovery involves integrating data from disparate sources like protein-ligand interactions, patent literature, genomics and clinical trials. Today, advancements in augmented reality (AR), virtual reality (VR), and high-performance computing (HPC) offer the tantalizing possibility of a similar interface revolutionizing the way scientists interact with multifactorial biology and chemistry datasets.

This post explores what a Minority Report-style interface for drug design would look like, how the seeds of such a system already exist in current technology, and the exciting potential this kind of interface holds for the future of drug discovery.

The Haptic, Gestural Future of Drug Discovery

Perhaps one of the most memorable aspects of Minority Report is the graceful, fluid way in which Tom Cruise’s character interacts with a futuristic interface using only his hands. With a series of quick, intuitive gestures, he navigates through complex data sets, zooming in on images, isolating key pieces of information, and piecing together the puzzle at the center of the plot. The thrill of this interface comes from its speed, accessibility, and above all, its elegance. Unlike the clunky, keyboard-and-mouse-driven systems we’re used to today, this interface allows data to be accessed and manipulated as effortlessly as waving a hand.

In drug discovery, such fluid navigation would be game-changing. As mentioned above, the modern scientist deals with a staggering amount of information: genomics data, chemical structures, protein-ligand interactions, toxicity reports, and clinical trial results, all coming from different sources. The ability to sweep through these datasets with a flick of the wrist, pulling in relevant data and discarding irrelevant noise in real-time, would make the process of drug design not only more efficient but more dynamic and creative. Imagine pulling together protein folding simulations, molecular docking results, and clinical trial metadata into a single, interactive, 3D workspace—all by making precise, intuitive hand movements like Tom Cruise.

The core of the Minority Report interface is its gestural and haptic nature, which would be crucial for translating such a UI into the realm of drug design. By introducing haptic feedback into the system—using vibrations or resistance in the air to simulate touch—a researcher could "feel" molecular structures, turning abstract chemical properties into tactile sensations. Imagine "grabbing" a molecule and feeling the properties of its surface—areas of hydrophobicity, polarity, or charge density—all while rotating the structure in mid-air with a flick of your wrist. Like an octopus sensing multiple inputs simultaneously, a researcher would be the active purveyor of a live datastream of multilayered data. This tactile feedback could become a new form of data visualization, where chemists and biologists no longer rely solely on charts and numbers but also on physical sensations to understand molecular behavior. The experience would translate to an entirely new dimension of interacting with molecular data and models, making it possible to “sense” molecular conformations in ways that are impossible with current 2D screens.

Such a haptic interface would also make the process more accessible. Students and new researchers in drug discovery would quickly learn how to navigate and manipulate datasets through a gestural UI. The muscle memory developed through these natural, human movements would make the learning curve less steep, transforming the learning experience into something more akin to a hands-on laboratory session rather than an abstract, numbers-on-a-screen challenge. Drug discovery and molecular design would be democratized.

Swiping Through Multifactorial Datasets

One of the most exciting possibilities of a Minority Report-style UI in drug discovery is its ability to merge multifactorial datasets, making complex biology and chemistry data "talk" to each other. In drug discovery, researchers deal with data from various domains, — genomics, proteomics, cheminformatics, clinical data etc. — each of which exists in its own silo; any researcher in the area would relate to the pain of integrating these very different databases, an endeavor that requires a significant amount of effort and specialized software. Currently, entire IT departments are employed to these ends. A futuristic UI could change that entirely.

Imagine a scientist swiping through an assay dataset with one hand, while simultaneously bringing in chemical structure data and purification data on stereoisomers with the other. Perhaps throw in a key blocking patent and gene expression data. These diverse datasets could then be overlaid in real time, with machine learning algorithms providing instant insights into correlations and potential drug candidates. For instance, one swipe could summon a heat map of gene expression related to a disease, while another flick could display how a particular small molecule binds to a target protein implicated in that disease. A few more gestures could allow the scientist to access historical drug trials and toxicity data as well as patent data, immediately seeing if any patterns emerge. The potential here is enormous: combining these multifactorial datasets in such a seamless, visual way would enable researchers to generate hypotheses on the fly, test molecular interactions in real-time, and identify the most promising drug candidates faster than ever before.

The Seeds Are Already Here: AR, VR, and High-Performance Computing

While this vision seems futuristic, the seeds of this interface already exist in today's technology. Augmented reality (AR) and virtual reality (VR) platforms are rapidly advancing, providing immersive environments that allow users to interact with data in three dimensions. AR devices like Microsoft's HoloLens and VR systems like the Oculus Rift already provide glimpses of what a 3D drug discovery workspace might look like. For example, AR could be used to visualize molecular structures in real space, allowing researchers to walk around a protein or zoom in on a ligand-binding site as if it were floating right in front of them.

At the same time, high-performance computing (HPC) is already pushing the limits of what we can do with drug discovery. Cloud-based platforms provide immense computing power that can process large datasets, while AI-driven software accelerates the pace of molecular docking simulations and virtual screening processes. Combining these technologies with a Minority Report-style interface could be the key to fully realizing the potential of this future workspace.

LLMs as Intelligent Assistants

While the immersive interface and tactile data manipulation are powerful, the addition of large language models (LLMs) brings an entirely new layer of intelligence to the equation. In this vision of drug discovery, LLMs would serve as intelligent research assistants, capable of understanding complex natural language queries and providing context-sensitive insights. Instead of manually pulling in data or running simulations, researchers could ask questions in natural language, and the LLM would retrieve relevant datasets, compute compound properties, run analyses, and even suggest possible next steps. Even if a researcher could summon up multiple datasets by swiping in an interactive display, they would still need an LLM to answer questions pertaining to cross-correlations between these datasets.

Imagine a researcher standing in front of an immersive display, surrounded by 3D visualizations of molecular structures and genomic data. With a simple voice command or text prompt, they could ask the LLM, “Which compounds have shown the most promise in targeting this specific binding site?” or “What genetic mutations are correlated with resistance to this drug?” or even fuzzier questions like “What is the probability that this compound would bind to the site and cause side effects?”. The LLM would then comb through millions of datasets, both existing and computed, and instantly provide answers, suggest hypotheses, or even propose new drug candidates based on historical data.

Moreover, LLMs could help interpret complex, multifactorial relationships between datasets. For example, if a researcher wanted to understand how a particular chemical compound might interact with a genetic mutation in cancer cells, they could ask the LLM to cross-reference all available data on drug resistance, molecular pathways, and previous clinical trials. The LLM could provide a detailed, synthesized response, saving the researcher countless hours of manual research and allowing them to focus on making creative, strategic decisions.

This kind of interaction would fundamentally change the way scientists approach drug discovery. No longer would they need to rely solely on their own ability to manually search for and interpret data. Instead, they could work in tandem with an intelligent, AI-driven system that helps them navigate the immense complexity of modern drug design. With the right interface, researchers could manipulate massive amounts of drug discovery data in real-time, powered by already existing HPC infrastructure.

Current challenges

While this vision of an all-in-one molecular design interface sounds promising, we would be remiss in not mentioning some familiar current challenges. Data is still highly siloed, even within organizations, and inter-organizational data sharing is still bound by significant legal, business and technological challenges. While AR and VR are now being democratized through increasingly cheap headsets and software, the experience is not as smooth as we would like, and bringing in disparate data sources into the user experience remains a problem. In the future, common API formats could become a game changer. Finally, LLMs still suffer from errors and hallucinations. Having a human in the loop would be imperative in overcoming these limitations, but there is little doubt that the sheer time-saving and consolidation they enable, along with the ability to query data in natural language, would make their use not just important but inevitable.

A Future of Instant, Integrated Data at Your Fingertips

The promise of a Minority Report-style interface for drug discovery lies in its ability to make data instantly accessible, integrated, and actionable. By swiping and gesturing in mid-air, scientists would no longer be constrained by traditional input methods, unlocking new levels of creativity and efficiency. This kind of interface would enable instant access to everything from raw molecular data to advanced machine-learning models predicting the efficacy of new drug candidates.

We can image a future where a drug designer could pull up decades of research on a specific disease, instantly overlay that with genomic data, and compare it with molecular screening results—all in a 3D, immersive environment. The heightened experience would make it possible to come up with radically new hypotheses about target engagement, efficacy and toxicity in short order. Collaboration would also reach new heights, as teams across the world interacted in the same virtual workspace, manipulating the same data sets in real time, regardless of their physical location. The interface would enable instant brainstorming, rapid hypothesis generation and testing, and seamless sharing of insights. The excitement surrounding such a future is palpable. By blending AR, VR, HPC, and LLMs, we can transform drug discovery into an immersive, highly interactive, and profoundly intuitive process.

Let the symphony start playing.

Areopagitica and the problem of regulating AI

How do we regulate a revolutionary new technology with great potential for harm and good? A 380-year-old polemic provides guidance.

In 1644, John Milton sat down to give a speech to the English parliament arguing in favor of the unlicensed printing of books and against a proposed bill to restrict their contents. Published as “Areopagitica”, Milton’s speech became one of the most brilliant defenses of free expression.

Milton rightly recognized the great potential books had and the dangers of smothering that potential before they were published. He did not mince words:

“For books are not absolutely dead things, but …do preserve as in a vial the purest efficacy and extraction of that living intellect that bred them. I know they are as lively, and as vigorously productive, as those fabulous Dragon’s teeth; and being sown up and down, may chance to spring up armed men….Yet on the other hand unless wariness be used, as good almost kill a Man as kill a good Book; who kills a Man kills a reasonable creature, God’s Image; but he who destroys a good Book, kills reason itself, kills the Image of God, as it were in the eye. Many a man lives a burden to the Earth; but a good Book is the precious life-blood of a master-spirit, embalmed and treasured up on purpose to a life beyond life.”

Apart from stifling free expression, the fundamental problem of regulation as Milton presciently recognized is that the good effects of any technology cannot be cleanly separated from the bad effects; every technology is what we call dual-use. Referring back all the way to Genesis and original sin, Milton said:

“Good and evil we know in the field of this world grow up together almost inseparably; and the knowledge of good is so involved and interwoven with the knowledge of evil, and in so many cunning resemblances hardly to be discerned, that those confused seeds which were imposed upon Psyche as an incessant labour to cull out, and sort asunder, were not intermixed. It was from out the rind of one apple tasted, that the knowledge of good and evil, as two twins cleaving together, leaped forth into the world.”

In important ways, “Areopagitica” is a blueprint for controlling potentially destructive modern technologies. Freeman Dyson applied the argument to propose commonsense legislation in the field of recombinant DNA technology. And today, I think, the argument applies cogently to AI.

AI is such a new technology that its benefits and harms are largely unknown and hard to distinguish from each other. In some cases the distinction is clear. For instance, image recognition can be used for all kinds of useful applications ranging from weather assessment to cancer cell analysis, but it can be and is used for surveillance. In that case, it is not possible to separate out the good from the bad even when we know what they are. But more importantly, as the technology of image recognition AI demonstrates, it is impossible to know what exactly AI will be used for unless there’s an opportunity to see some real-world applications of it. Restricting AI before these applications are known will almost certainly ensure that the good applications are stamped out.

It is in the context of Areopagitica and the inherent difficulty of regulating a technology before its potential is unknown that I find myself concerned about some of the new government regulation which is being proposed for regulating AI, especially California Bill SB-1047 which has already passed the state Senate and has made its way to the Assembly, with a proposed decision date at the end of this month.

The bill proposes commonsense measures for AI, such as more transparent cost-accounting and documentation. But it also imposes what seem like arbitrary restrictions on AI models. For instance, it would require regulation and paperwork for models which cost $100 million or more per training run. While this regulation will exempt companies which run cheaper models, the problem in fact runs the other way: nothing stops cheaper models from being used for nefarious purposes.

Let’s take a concrete example: in the field of chemical synthesis, AI models are increasingly used to do what is called retrosynthesis, which is to virtually break down a complex molecule into its constituent building blocks and raw materials (as a simple example, a breakdown of sodium chloride into sodium and chlorine would be retrosynthesis). One can use retrosynthesis algorithms to find out the cheapest or the most environmentally friendly route to a target molecule like a drug, a pesticide or an energy material. And run in reverse, you can use the algorithm for forward planning, predicting based on building blocks what the resulting target molecule would look like. But nothing stops the algorithm from doing the same analysis on a nerve gas or a paralytic or an explosive; it’s the same science and the same code. Importantly, much of this analysis is now available in the form of laptop computer software which enables the models to be trained on datasets of millions of data points: small potatoes in the world of AI. Almost none of these models cost anywhere close to $100 million, which puts their use in the hands of small businesses, graduate students and – if and when they choose to use them – malicious state and non-state actors.

Thus, restricting AI regulation to expensive models might exempt smaller actors, but it’s precisely that fact that would enable these small actors to use the technology to bad ends. On the other hand, critics are also right that it would effectively price out the good small actors since they would not be able to afford the legal paperwork that the bigger corporations can. The arbitrary cap of $100 million therefore does not seem to address the root of the problem. The same issue applies to another restriction which is also part of the European AI regulation, which is limiting the calculation speed to 1026 flops. Using the same example of the AI retrosynthesis models, it is easy to argue that such models can be run for far less computing power and would still produce useful results.

What then is the correct way to regulate AI technology? Quite apart from the details, one thing that is clear is that we should be able to experiment a bit, run laboratory-scale models and at least try to probe the boundaries of potential risks before we decide to stifle this or that model or rein in computing power. Once again Milton echoes such sentiments. As a 17th century intellectual it would have been a long shot for him to call for the completely free dissemination of knowledge; he must well have been aware of the blood that had been shed in religious conflicts in Europe during his time. Instead, he proposed that there could be some checks and restrictions on books, but only after they had been published:

“If then the Order shall not be vain and frustrate, behold a new labour, Lords and Commons, ye must repeal and proscribe all scandalous and unlicensed books already printed and divulged; after ye have drawn them up into a list, that all may know which are condemned, and which not.”

Thus, Milton was arguing that books should not be stifled at the time of their creation; instead, they should be stifled at the time of their use if the censors saw a need. The creation vs use distinction is a sensible one when thinking about regulating AI as well. But even that distinction doesn’t completely address the issue, since the uses of AI technology are myriad, and most of them are going to be beneficial and intrinsically dual-use. Even regulating the uses of AI thus would entail interfering in many aspects of AI development and deployment. And what about the legal and commercial paperwork, the extensive regulatory framework and the army of bureaucrats that would be needed to enforce this legislation? The problem with legislation is that it is easy for it to overstep boundaries, to be on a slippery slope and gradually elbow its way into all kinds of things for which it wasn’t originally intended, exceeding its original mandate. Milton shrewdly recognized this overreach when he asked what else besides printing might be up for regulation:

“If we think to regulate printing, thereby to rectify manners, we must regulate all recreations and pastimes, all that is delightful to man. No music must be heard, no song be set or sung, but what is grave and Doric. There must be licensing dancers, that no gesture, motion, or deportment be taught our youth but what by their allowance shall be thought honest; for such Plato was provided of; it will ask more than the work of twenty licensers to examine all the lutes, the violins, and the guitars in every house; they must not be suffered to prattle as they do, but must be licensed what they may say. And who shall silence all the airs and madrigals that whisper softness in chambers? The windows also, and the balconies must be thought on; there are shrewd books, with dangerous frontispieces, set to sale; who shall prohibit them, shall twenty licensers?”

This passage shows that not only was John Milton a great writer and polemicist, but he also had a fine sense of humor. Areopagitica shows us that if we are to confront the problem of AI legislation, we must do it not just with good sense but with a recognition of the absurdities which too much regulation may bring.

The proponents of AI who fear the many problems it might create are well-meaning, but they are unduly adhering to the Precautionary Principle. The Precautionary Principles says that it’s sensible to regulate something when its risks are not known. I would like to suggest that we replace the Precautionary Principle with a principle I call The Adventure Principle. The Adventure Principle says that we should embrace risks rather than running away from them because of the benefits which exploration brings. Without the Adventure Principle, Columbus, Cook, Heyerdahl and Armstrong would never have set sail into the great unknown and Edison, Jobs, Gates and Musk would never embark on big technological projects. Just like with AI, these explorers faced a significant risk of death and destruction, but they understood that with immense risks come immense benefits, and by the rational light of science and calculation, they thought there was a good chance that the risks could be managed. They were right.

Ultimately there is no foolproof “pre-release” legislation or restriction that would purely stop the bad use of models while still enabling their good use. Milton’s Areopagitica does not tell us what the right legislation for regulating AI would look like, although it provides hints based on regulation of use rather than creation. But it makes a resounding case for the problems that such legislation may create. Regulating AI before we have a chance to see what it can do would be like imprisoning a child before he grows up into a young man. Perhaps a better approach would be the one Faraday adopted when Gladstone purportedly asked him what the use of electricity was: “Someday you may tax it”, was Faraday’s response.

Some say that the potential risks from AI are too great to allow for such a liberal approach. But the potential risks from almost any groundbreaking technology developed in the last few hundred decades – printing, electricity, fossil fuels, automobiles, nuclear energy, gene editing – are no different. The premature regulation of AI would prevent us from unleashing its potential to confront our most pressing challenges. When humanity is then grasping with its last-ditch efforts to prevent its own extinction because of known problems, a recognition of the irony of smothering AI because of a fear of unknown problems would come too late to save us.

Book Review: Chip War: The Fight for the World's Most Critical Technology

In the 19th century it was coal and steel, in the 20th century it was oil and gas, what will it be in the 21st century? The answer, according to Chris Miller in this lively and sweeping book, is semiconductor chips.

There is little doubt that chips are ubiquitous, not just in our computer and cell phones but in our washers and dryers, our dishwashers and ovens, our cars and sprinklers, in hospital monitors and security systems, in rockets and military drones. Modern life as we know it would be unimaginable without these marvels of silicon and germanium. And as Miller describes, we have a problem because much of the technology to make these existential entities is the province of a handful of companies and countries that are caught in geopolitical conflict.

Miller starts by tracing out the arc of the semiconductor industry and its growth in the United States, driven by pioneers like William Shockley, Andy Grove and Gordon Moore and fueled by demands from the defense establishment during the Cold War. Moore's Law has guaranteed that the demand and supply for chips has exploded in the last few decades; pronouncements of its decline have often been premature. Miller also talks about little-known but critically important people like Weldon Ward who designed chips that made precision missiles and weapons possible, secretary of defense Bill Perry who pressed the Pentagon for funding and developing precision weapons and Lynn Conway, a transgender scientist who laid the foundations for chip design.

Weldon Ward's early design for a precision guided missile in Vietnam was particularly neat: a small window in the tip of the warhead shined laser back to a chip that was divided into four quadrants. If one quadrant started getting more light than the other you would know the missile was off-course and would adjust it. Before he designed the missile, Ward was shown photos of a bridge in Vietnam that was surrounded by craters that indicated where the missile had hit. After he designed his missile, there were no more craters, only a destroyed bridge.

There are three kinds of chips: memory chips which control the RAM in your computer, logic chips which control the CPU and analog chips which control things like temperature and pressure sensing in appliances. While much of the pioneering work in designing transistors and chips was spearheaded by American scientists at companies like Intel and Texas Instruments, soon the landscape shifted. First the Japanese led by Sony's Akio Morita captured the market for memory or DRAM chips in the 80s before Andy Grove powerfully brought it back to the US by foreseeing the personal computer era and retooling Intel for making laptop chips. The landscape also shifted because the U.S. found cheap labor in Asia and outsourced much of the manufacturing of chips.

But the real major player in this shift was Morris Chang. Chang was one of the early employees at Texas Instruments and his speciality was in optimizing the chemical and industrial processes for yielding high-quality silicon. He rose through the ranks and advised the defense department. But, in one of those momentous quirks of history that at the time sound trivial, he was passed over for the CEO position. Fortunately he found a receptive audience in the Taiwanese government who gave him a no-strings-attached opportunity to set up a chip manufacturing plant in Taiwan.

The resulting company, TSMC, has been both the boon and the bane of the electronics age. If you use a device with a chip in it, it has most probably been made by TSMC. Apple, Amazon, Tesla, Intel, all design their own chips but have them made by TSMC. However it does not help that TSMC is located in a company that both sits on top of a major earthquake fault and is the target for invasion or takeover by a gigantic world power. The question of whether our modern technology that is dependent on chips can thrive is closely related to whether China is going to invade Taiwan.

The rest of the supply chain for making chips is equally far flung. But although it sounds globalized, it's not. For instance the stunningly sophisticated process of extreme ultraviolet lithography (EUV) that etches designs on chips is essentially monopolized by one company - ASML in the Netherlands. The machines to do this cost more than $100 million each and have about 500,000 moving parts. If something were to happen to ASML the world's chip supply would come to a grinding halt.

The same goes for the companies that make the software for designing the chips. Three companies in particular - Cadence, Synopsys and Mentor - make 90% of chip design software. There are a handful of other companies making specialized software and hardware, but they are all narrowly located.

Miller makes the argument that the future of chips, and therefore of modern technology at large, is going to depend on the geopolitical relationship especially between China and the United States. The good news is that currently China lags significantly behind the U.S. in almost all aspects of chip design and manufacturing; the major centers for these processes are either in the U.S. or in countries which are allies of the U.S. In addition, replicating machinery of the kind used for etching by ASML is hideously complicated. The bad news is that China has a lot of smart scientists and engineers and uses theft and deception to gain access to chip design and making technology. Using front companies and legitimate buyouts, they have already tried to gain such access. While it will still take years for them to catch up, it is more a question of when than if.

If we are to continue our modern way of life that depends on this critical technology, it will have to be done through multiple fronts, some of which are already being set in motion. Intel is now setting up its own foundry and trying to replicate some of the technology that ASML uses. China will have to be brought to the bargaining table and every attempt will have to be made to ensure that they play fair.

But much of the progress also depends on funding basic science. It's worth remembering that much of the early pioneering work in semiconductors was done by physicists and chemists at places like Bell Labs and Intel, a lot of it by immigrants like Andy Grove and Morris Chang. Basic research at national labs like Los Alamos and Sandia laid the foundations for ASML's etching technology. Attempts to circumvent Moore's Law will also have to be continued to be made; as transistors shrink down to single digit nanometer sizes, quantum effects make their functioning more uncertain. However there are plans to avoid these issues through strategies like stacking them together. All these strategies depend on training the next generation of scientists and engineers, because progress on technology ultimately depends on education.

Philip Morrison on challenges with AI

Philip Morrison who was a top-notch physicist and polymath with an incredible knowledge of things beyond his immediate field was also a speed reader who reviewed hundreds of books on a stunning range of topics. In one of his essays from an essay collection he held forth on what he thought were the significant challenges with machine intelligence. It strikes me that many of these are still valid (italics mine).

"First, a machine simulating the human mind can have no simple optimization game it wants to play, no single function to maximize in its decision making, because one urge to optimize counts for little until it is surrounded by many conditions. A whole set of vectors must be optimized at once. And under some circumstances, they will conflict, and the machine that simulates life will have the whole problem of the conflicting motive, which we know well in ourselves and in all our literature.

Second, probably less essential, the machine will likely require a multisensory kind of input and output in dealing with the world. It is not utterly essential, because we know a few heroic people, say, Helen Keller-who managed with a very modest cross-sensory connection nevertheless to depict the world in some fashion. It was very difficult, for it is the cross-linking of different senses which counts. Even in astronomy, if something is "seen" by radio and by optics, one begins to know what it is. If you do not "see" it in more than one way, you are not very clear what it in fact is.

Third, people have to be active. I do not think a merely passive machine, which simply reads the program it is given, or hears the input, or receives a memory file, can possibly be enough to simulate the human mind. It must try experiments like those we constantly try in childhood unthinkingly, but instructed by built-in mechanisms. It must try to arrange the world in different fashions.

Fourth, I do not think it can be individual. It must be social in nature. It must accumulate the work--the languages, if you will- of other machines with wide experience. While human beings might be regarded collectively as general-purpose devices, individually they do not impress me much that way at all. Every day I meet people who know things I could not possibly know and can do things I could not possibly do, not because we are from differing species, not because we have different machine natures, but because we have been programmed differently by a variety of experiences as well as by individual genetic legacies. I strongly suspect that this phenomenon will reappear in machines that specialize, and then share experiences with one another. A mathematical theorem of Turing tells us that there is an equivalence in that one machine's talents can be transformed mathematically to another's. This gives us a kind of guarantee of unity in the world, but there is a wide difference between that unity, and a choice among possible domains of activity. I suspect that machines will have that choice, too. The absence of a general-purpose mind in humans reflects the importance of history and of development. Machines, if they are to simulate this behavior- or as I prefer to say, share it--must grow inwardly diversified, and outwardly sociable.

Fifth, it must have a history as a species, an evolution. It cannot be born like Athena, from the head full-blown. It will have an archaeological and probably a sequential development from its ancestors. This appears possible. Here is one of computer science's slogans, influenced by the early rise of molecular microbiology: A tape, a machine whose instructions are encoded on the tape, and a copying machine. The three describe together a self-reproducing structure. This is a liberating slogan; it was meant to solve a problem in logic, and I think it did, for all but the professional logicians. The problem is one of the infinite regress which looms when a machine becomes competent enough to reproduce itself. Must it then be more complicated than itself? Nonsense soon follows. A very long

instruction tape and a complex but finite machine that works on those instructions is the solution to the logical problem."

A potentially revolutionary new technique for chemical structure determination

I am a big believer in science as a tool-driven rather than an idea-driven revolution, and nowhere do you see this view of science exemplified better than in the development of instrumental techniques in chemistry - most notably NMR and x-ray diffraction. NMR and crystallography were not just better ways to see molecules, but in their scope, their throughput, their cost and their speed, they opened up whole new fields of science like genomics and nanomaterials up to investigation that their developers couldn't even have imagined.

So it is with some interest that I saw a paper from the Nelson, Gonen and Rodriguez labs at UCLA and the Stoltz lab at Caltech that describes a new way to rapidly determine the structures of organic molecules. And I have to say: very few papers in recent times have made me sit up and do a double take, but this one did. At one point in the paper the authors say they were "astounded" by the ease of the technique, and I don't think that word is out of place here at all.

Until now, crystallography has been the gold standard for all kinds of structure determination; it gives you as direct a view of molecules at the atomic level as possible. But the very name "crystallography" implies that you need to get your sample into a crystalline state, and as any chemist who has worked with a headache-inducing list of assorted powders, gels, oils and tars knows, being crystalline is not the natural state of most molecules. In most cases your samples are thus simply not in a convenient form for crystallography.

That's why NMR has been the primary technique for routine organic structure determination. But NMR is still relatively slow and depends on having a machine that's expensive and sometimes breaks down. You also cannot do NMR on a benchtop, and getting the sample in a pure enough condition in the right solvent is also key to good structure determination. Then there are all the problems attendant with shimming, water suppression and other artifacts that NMR presents, although sophisticated software can now take care of most of these. Nonetheless, as powerful as NMR is and will continue to be, it's not exactly a rapid, plug-and-play system.

That's why this recent paper is so promising. It describes a crystallographic technique that uses cryo-EM and micro-electron diffraction (micro ED) for efficiently finding out the structure of organic molecules. Electron diffraction itself is an old technique, pioneered for instance by Linus Pauling in the 1930s, but this is not any old ED, it's micro ED. Cryo-EM already won the Nobel Prize two years ago for determining the structures of complex proteins, but it has never been routinely applied to small molecule structure determination. This new technique could change that landscape in a jiffy. And I mean in a jiffy - the examples they have shown take a few minutes each. The first molecule - progesterone - went from powder to pattern in less than 30 mins, which is quite stunning. And the resolution was 1 Å, and you can't ask for more. Up to twelve samples were investigated in a single experiment.

But what really made me sit up was the variety of starting points that could be investigated. From amorphous powders to samples straight out of flash chromatography to mixtures of compounds, the method made quick work out of everything. As mentioned above, amorphous powders and mixtures are the rule rather than the exception in standard organic synthesis, so one can see this technique being applied to almost every chemical purification and synthetic manipulation done in routine synthesis or structure determination. For me the most amazing application however was the determination of a mixture of four different molecules: no other technique in organic chemistry which I know can do mixtures in a few minutes with such high resolution with such little material.

There are undoubtedly still limitations. For one thing, cryo-electron microscopes are still not cheap, and while sample preparation is getting better, it's also not instantaneous in every case. I also noticed that most of the compounds this study looked at were rather rigid, with lots of fused and other rings; floppy molecules will likely cause some trouble, and although thiostrepton is an impressive-looking molecule, it would be interesting to see how this works for beasts like rapamycin or oligopeptides. In general, as with other promising techniques, we will have to see what the domain of applicability of this method is. 

Nonetheless, this is the kind of technique that promises to take a scientific field in very novel directions. It could accelerate the everyday practice of organic chemistry in multiple fields - natural products, chemical biology, materials science - many fold; and at some point, quantity has a quality of its own. It could allow the investigation of the vast majority of compounds that cannot be easily coaxed into a crystal or an NMR tube. And it could perhaps even allow us to study conformational behavior of floppy compounds, which from first-hand experience I know is pretty hard to do.

If validated, this technique also exemplifies something I have talked about before, which is how scientific tools and discoveries build on each other; in other words, how scientific convergence is a key driving force in science. When cryogenics was invented, nobody foresaw cryo-electron microscopy, and when cryo-EM was invented, nobody foresaw its application to routine organic synthesis. And so it goes on, science and technology piggybacking in ever-expanding spirals.

Technological convergence in drug discovery and other endeavors

You would think that the Wright brothers’ historic flight from Kitty Hawk on December 17, 1903 had little to do with chemistry. And yet it did. The engine they used came from an aluminum mold; since then aluminum has been a crucial ingredient in lightweight flying machines. The aluminum mold would not have been possible had industrial chemists like Charles Hall and Paul Héroult not developed processes like the Hall-Héroult process for refining the metal from its ore, bauxite. More elementally, the gasoline fueling the flight was the result of a refining process invented more than fifty years earlier by a Yale chemist named Benjamin Silliman. There was a fairly straight line from the Bayer and Silliman processes to Kitty Hawk.

The story of the Wright brothers’ powered flight illustrates the critical phenomenon of technological convergence that underlies all major technological developments in world history. Simply put, technological convergence refers to the fact that several enabling technologies have to come together in order for a specific overarching technology to work. And yet what’s often seen is only the technology that benefits, not the technology that enables.

We see technological convergence everywhere. Just to take a few of the most important innovations of the last two hundred years or so: The computer would not have been possible without the twin inventions of the transistor and silicon purification. MRI would not have been possible without the development of sophisticated software to deconvolute magnetic resonance signals and powerful magnets to observe those signals in the first place. There are other important global inventions that we take for granted - factory farming, made-to-order houses, fiber optics, even new tools like machine learning - none of which would have materialized had it not been for ancillary technologies which had to reach maturation.

Recognizing technological convergence is important, both because it helps us appreciate how much has to happen before a particular technology can embed itself in people’s everyday lives, and because it can help us potentially recognize multiple threads of innovation that could potentially converge in the future - a risky but important vision that can help innovators and businessmen stay ahead of the curve. One important point to note: by no means does technological convergence itself help innovations rise to the top – political and social factors can be as or more crucial – but this convergence is often necessary even if not sufficient.

It’s interesting to think of technological convergence in my own field of drug development. Let’s look at a few innovations, both more recent as well as older, that illustrate the phenomenon. Take a well-established technology like high-throughput screening (HTS). HTS came on the scene about thirty years ago, and since then has contributed significantly to the discovery of new medicines. What made the efficient screening of tens of thousands of compounds possible? Several convergent developments: recombinant DNA technology for obtaining reasonable quantities of pure proteins for screening, robotic techniques and automation for testing these compounds quickly at well-defined concentrations in multiple wells or plates, spectroscopic techniques like FRET for determining the feasibility of the end results, and graphing and visualization software for mapping the results and quickly judging if they made sense. These are just a few developments: in addition, there are techniques within these techniques that were also critical. For instance, recombinant DNA depended on methods for viral transfection, for splicing and ligation and for sequencing, and robotic automation depended on microelectronic control systems and materials for smooth manipulation of robotic moving parts. Thus, not only is technology convergent but it also piggybacks, with one piece of technology building on another to produce a whole that is more than the sum of its parts, aiding in the success of a technology it wasn’t primarily designed for.

Below is a table of just a few other primary drug discovery technologies that could not have been possible without ancillary convergent technologies.

Primary technology	Convergent enabling technologies
Combinatorial chemistry	LCMS for purification, organic synthesis methodology, hardware (solid phase beads, plastic, tubes, glassware) for separation and bookkeeping.
Molecular modeling	Computing power (CPUs, GPUs), visualization software, crystal structures and databases (PDB, CSD etc.)
Directed evolution/phage display	Recombinant DNA technology, hardware (solid phase supports), buffer chemistry for elution.
DNA-encoded libraries	PCR, DNA sequencing technology (Illumina etc.), hardware (solid phase beads, micropipettes etc.), informatics software for deconvolution of results.
NMR	Cryogenics, magnet production, software.

I have deliberately included NMR spectroscopy in the last row. A modern day organic chemist’s work would be unthinkable without this technique. It of course depends crucially on the availability of high-field magnets and the cryogenics techniques that keep the magnet cold by immersion in liquid helium, but it also depends fundamentally on the physics of nuclear magnets worked out by Isidor Rabi, Edward Purcell, Richard Ernst and others. Since this post is about technology I won’t say anything further about science, but it should be obvious that every major technology rests on a foundation of pure science which has to be developed for decades before it can be applied, often with no clear goal in mind. Sometimes the application can be very quick, however. For instance, it’s not an accident that solid phase supports appear in three of the five innovations listed above. Bruce Merrifield won the Nobel Prize in chemistry for his development of solid-phase peptide synthesis in 1984, and a little more than thirty years later, that development has impacted many enabling drug development techniques.

There are two interesting conclusions that emerge from considering technological convergence. The first is the depressing conclusion that if ancillary technologies haven’t kept pace, then even the most brilliant innovative idea would get nowhere. Even the most perspicacious inventor won’t be able to make a dent in the technology universe, simply because the rest of technology hasn’t kept up with him. A good example is the early spate of mobile phones appearing in the early 90s which didn’t go anywhere. Not only were they too expensive, but they simply weren’t ready for prime time because the wide availability of broadband internet, touchscreens and advanced battery technology was non-existent. Similarly, the iPhone and iPod took off not just because of Steve Jobs’ sales skills and their sleek GUI, but because broadband internet, mp3s (both legal and pirated) and advanced lithium ion batteries were now available for mass production. In fact, the iPod and the iPhone showcase convergent technologies in another interesting way; their sales skyrocketed because of the iTunes Music Store and the iPhone App store. As the story goes, Jobs was not sold on the app store idea for a long time because he characteristically wanted to keep iPhone apps exclusive. It was only flagging initial sales combined with insistent prodding from the iPhone team that changed his mind. In this case, therefore, the true convergent technology was not really battery chemistry or the accelerometer in the phone but a simple software innovation and a website.

The more positive conclusion to be drawn from the story of convergent technology is to keep track of ancillary enabling technologies if you want to stay ahead of the curve. In case of the iPod, Jobs seems to have had the patience to wait before USB, battery and internet technologies became mature enough for Apple to release the device; in spite of being the third or fourth mp3 player on the market, the iPod virtually took over in a few years. What this means for innovators and technologists is that they should keep an eye out on the ‘fringe’, on seemingly minor details of their idea that might have a crucial impact on its development or lack thereof. If you try to launch an innovative product before the ancillary technologies have caught up, you won’t achieve convergence and the product might well be doomed.

Of course, groundbreaking ancillary technologies are often obvious only in retrospect and are unexpected when they appear – Xerox’s mouse and GUI come to mind – but that does not mean they are invisible. One reason John D. Rockefeller became so spectacularly successful and wealthy is because he looked around the corner and saw not one but three key technologies: oil drilling, oil transportation and oil refining. Similarly, Edison’s success owed, in part, to the fact that he was an all-rounder, developing everything from electrical circuits to the right materials for bulb filaments; chemistry, electricity, mechanical engineering – all found a home in Edison’s lab. Thus, while it’s not guaranteed, one formula for noting the presence or absence of technological convergence is to cast a wide net, to work the field as well as its corners, to spend serious time exploring even the small parts that are expected to contribute to the whole. Recognizing technological convergence requires a can-do attitude and the enthusiasm to look everywhere for every possible lead.

At the very least, being cognizant of convergent technologies can prevent us from wasting time and effort; for instance, combinatorial chemistry went nowhere at the beginning because HTS was not developed. Molecular modeling went nowhere because sampling and scoring weren’t well developed. Genome sequencing by itself went nowhere because simply having a list of genes rang hollow until the technologies for interrogating their protein products and functions weren’t equally efficient. Developing your technology in a silo, no matter how promising it looks by itself, can be a failing effort if not fortified with other developing technology which you should be on the lookout for.

Technology, like life on earth, is part of an ecosystem. Even breakthrough technology does not develop in a vacuum. Without convergence between different innovations, every piece of technology would be stillborn. Without the aluminum, without the refined petroleum, the Wright Flyer would have lain still in the sands of the Outer Banks.