Not all that long ago, the future seemed to have stalled. The last true generational shift in technology was the mobile web, but that was so old it had become merely “the web”. And as technologists sat patiently, waiting for the next big thing to come along, fever dreams about “the metaverse” and “Web3” filled the vacuum. The hottest conversations in technology were about things that didn’t actually exist.
How quickly things change.
Over the past year, generative AI has made astonishing, rapid progress, resetting technology’s center of gravity and raising profound questions for society. The future is once again firing on all cylinders—in fact it’s moving so fast it can be disorientating.
So, how to get your bearings at a time like this? When it’s hard to forecast what’s next, the best advice is to pay close attention to what’s actually happening at the frontier, in case you need to respond. “Usually these things aren’t brand new—they’re things that have evolved in some way that makes them more important, or means they’re having greater impact,” says Rachel Laycock, Chief Technology Officer at the technology consultancy Thoughtworks. “Even generative AI—it’s not new, it’s just more accessible and there are things that are new within the field.”
And for business leaders trying to stay on the front foot, remember there’s more happening out there than simply AI. Sure, that may be the megatrend of the moment, but there are other important developments that should also be on your horizon.
Here are three trends to track…
The old joke is that quantum computers are always a decade away. And sure, despite continual progress thanks to major investments by tech companies, the era of practical quantum computing is unlikely to dawn soon. But business leaders should be planning for when it does. Specifically, they need to be taking preemptive action against the security threat they pose—and 2024 is the year that can start in earnest.
First, a quick—and highly simplified—review. Classical computers perform calculations in bits that can be in a binary state of 0 or 1. Quantum computers, however, use “qubits”, which can represent both 0 and 1 at the same time. Combine that with another mysterious quality known as “quantum entanglement”, which allows qubits to have instantaneous effects on other qubits, and they can perform tasks much faster than even the most advanced supercomputers.
Troublingly for businesses, that also applies to breaking current encryption methods. “Most of the crypto systems that we use today will be completely vulnerable,” says Dustin Moody, who leads on post-quantum cryptography at the US National Institute of Standards and Technology (NIST). “If you’re using those systems, adversaries with a quantum computer will be able to get access to your data.”
And the threat is already here, in a sense. “Harvest now, decrypt later” attacks are where hackers steal and stash sensitive data until such a point in the future as they can break into it.
Hence the burgeoning field of post-quantum security. Since 2016, NIST has been developing quantum-resistant algorithms and is planning to make its initial group of them ready for use in the first half of this year. “Having these standards finally ready to go is a key moment,” says Moody. “People can start using them and putting them into their products.” He imagines that early adopters will be organizations with high value information that needs to be protected well into the future, such as financial institutions, critical infrastructure, and major tech companies.
Although NIST is based in the US, its work on post-quantum cryptography has influence far beyond. Countries such as the UK, France, and Germany, for example, are recommending national usage of the NIST’s algorithms. “But hopefully if you need to protect yourself, then you’re already starting this process now,” says Moody. “There’s going to need to be a lot of preparation and planning: Taking an inventory of your products and what cryptography they use; establishing what’s quantum-vulnerable; budgeting money for the transition and bringing expertise onto staff. You can get ready well in advance of the algorithms coming.”
The phrase “large language model” has moved from academic journals to mainstream discourse over the past year. But while we’ve all grown familiar with the concept (a deep learning model trained on vast amounts of text), and been astonished by their abilities (Write code! Draft emails!), it is an arena still defined more by questions than answers. A particularly prominent example: What are the non-obvious ways this will change how businesses operate—what new tools will LLMs provide that go beyond simply writing documents?
One emerging use case is chatbots for internal know-how. A company’s intellectual capital is immensely valuable but distributed across multiple files and systems. The value of drawing all that knowledge together in a way that preserves it and makes it searchable has long been recognized. In the 1990s and early 2000s, there was a trend for firms hiring “knowledge managers” to manually do just that. “But it required a lot of effort,” says Tom Davenport, the President’s Distinguished Professor of IT and Management at Babson College in Massachusetts, and a research fellow at the MIT Center for Digital Business. “So the knowledge management movement largely died.”
LLMs have put it back in the spotlight. While we might conventionally think of LLMs as models that have been trained on the colossal corpus of the internet, it’s possible to turn their language processing skills and general reasoning abilities on corporate documentation. This enables the model to respond to natural language questions about anything from financial strategy to project management.
There are a few different approaches to creating such a model. At one end of the spectrum, firms can build their own models from scratch, but this is an enormous undertaking. At the other is the method which is taking off right now: “retrieval-augmented generation”, which involves grounding an existing model on internal data to supplement what it knows.
In the past few months, a number of high-profile companies—mainly in professional and financial services—have done just this, providing their staff with expert chatbot coworkers that have internalized the company’s collective thinking. That trend is set to grow. “I think it will become much easier from an LLM perspective to load the content and access large volumes of content,” says Davenport. “That aspect of the technology is already rapidly maturing, and you’re already seeing new tools and products emerge to help with the technical challenges.”
While this makes it much easier for firms to create their own AI know-it-alls, there is still a major hurdle to overcome. “Unfortunately, it has not gotten any easier to curate the knowledge in the first place, and you have to be quite careful with what you put in if you want the answers to be relevant and have fewer errors,” says Davenport. And that could take us full circle. “Basically, I think it means we have to go back to having knowledge managers. Just ones with a different set of capabilities who understand all these new technologies that didn't exist for the last generation.”
Behind the scenes of commerce, a quiet revolution is underway.
Traditionally, if you wanted to pay for something without using cash, you reached for your credit or debit card, either in plastic or virtual form. These cards are controlled by a small number of companies who have enjoyed market dominance for years. But a series of changes to financial infrastructure have enabled the emergence of a new way to pay: Instant account-to-account (A2A) payments. These are set to boom this year, and—as they gain a stronger footing in ecommerce—start to evolve in ways that will change how businesses operate.
Sending money from your bank account to another person or business is a long-established feature of banking. But this can now happen much faster and more conveniently, making it a more everyday way to pay, thanks to two tech-driven trends.
First is Open Banking, which has mandated APIs that make it possible for consumers to initiate A2A payments over user-friendly, third-party interfaces. Second is the rise of instant payment networks (or “rails”, in the jargon) which reduce transaction times to under a few seconds. These have unleashed the power of A2A, particularly for person-to-person payments, and now they’re coming to consumer transactions. If you want to buy a good or service, it is becoming increasingly possible to leave your cards alone and “pay by bank” with an instant, Open Banking-driven A2A payment instead.
This has key benefits for sellers. “Quite often when you make a card payment, the merchant doesn't get the money immediately,” says Mark Jones, Product Manager at Pay.UK, which operates the UK’s national retail payment systems. “Today, if you’re a big retailer, you’ll probably get it the same day. Whereas if you’re a small corner shop, it could be a couple of days before you see that money. And those transactions also come with a cost, and it generally falls to the merchant to foot the bill for it. Open Banking payments cut out costs and wait times.” A2A Payments also offer benefits to consumers. In certain regions such as Europe, Open Banking-based transactions make it easier to transact across borders. It is potentially a more frictionless experience, too, because customers don’t have to fill out card details.
A2A payments are expected to grow rapidly this year, making particular inroads to ecommerce, with more businesses offering it as an option at checkout. “There is huge enthusiasm within the sector to push this forward as soon as possible, as it’s a big opportunity,” says Jones. “Currently, A2A payments are three percent of our total volume, but they’re growing really fast.” The trend is particularly expected to hit the US; the country recently launched its FedNow real-time payments network to complement the existing instant payments network from The Clearing House known as Real Time Payments (RTP), and its payments landscape is viewed as ripe for disruption.
As this idea matures, we may see a number of innovations come to the fore. One is new approaches to refunds. “The refund process is terrible at the moment. It's like the second-class post: it can take four or five days for money to get back to the customer,” says Jones. “Instant A2A payments can really change that.” The second is a new set of regulations to protect consumers and define liability in the event of fraudulent transactions.
All in, the disruption this implies for card operators is obvious, and they are already adapting. What’s more, if you thought those card networks would still have physical shops to themselves, well, don’t be so certain. “Physical retail is the big target here,” says Jones. He imagines a future where you can tap your phone to “pay by bank” at the checkout, just as you do with a virtual credit-card today. “We’ve got a couple of steps to go before we get there. One of those is that digital wallet providers are quite restrictive with what those wallets can do, so there needs to be conversations to open up those capabilities. But we are making progress. It’s a really exciting time.”
