Within the tech trade, 2023 was a 12 months of transformation.
Spurred by the success of final 12 months’s breakout tech star, ChatGPT, Silicon Valley’s giants rushed to show themselves into synthetic intelligence corporations, jamming generative A.I. options into their merchandise and racing to build their very own, extra highly effective A.I. fashions. They did so whereas navigating an unsure tech economic system, with layoffs and pivots galore, and whereas making an attempt to maintain their growing older enterprise fashions aloft.
Not every part went easily. There have been misbehaving chatbots, crypto foibles and bank failures. After which in November, ChatGPT’s maker, OpenAI, melted down (and rapidly reconstituted itself) over a failed boardroom coup, proving as soon as and for all that there’s no such factor in tech as resting in your laurels.
Each December in my Good Tech Awards column, I attempt to neutralize my very own negativity bias by highlighting a number of lesser-known tech initiatives that struck me as useful. This 12 months, as you’ll see, most of the awards contain synthetic intelligence, however my aim was to sidestep the polarized debates about whether or not A.I. will destroy the world or put it aside and as an alternative concentrate on the right here and now. What’s A.I. good for at present? Whom is it serving to? What sorts of vital breakthroughs are already being made with A.I. as a catalyst?
As all the time, my award standards are imprecise and subjective, and no precise trophies or prizes are concerned. These are simply small, private blurbs of appreciation for a number of tech initiatives I believed had actual, apparent worth to humanity in 2023.
To Be My Eyes, Apple and researchers on the College of Texas at Austin, for enhancing accessibility by way of A.I.
Accessibility — the time period for making tech merchandise extra usable by individuals with disabilities — has been an underappreciated space of enchancment this 12 months. A number of current advances in synthetic intelligence — similar to multimodal A.I. fashions that may interpret pictures and switch textual content into speech — have made it attainable for tech corporations to construct new options for disabled customers. That is, I’d argue, an unambiguously good use of A.I., and an space the place individuals’s lives are already enhancing in significant methods.
I requested Steven Aquino, a contract journalist who focuses on accessible tech, to advocate his prime accessibility breakthroughs of 2023. He really helpful Be My Eyes, an organization that makes know-how for individuals with impaired imaginative and prescient. In 2023, Be My Eyes introduced a feature known as Be My AI, powered by OpenAI’s know-how, that permits blind and low-sighted individuals to purpose their smartphone digicam at an object and have that object described for them in pure language.
Mr. Aquino additionally pointed me to Apple’s new Personal Voice feature, which is constructed into iOS 17 and makes use of A.I. voice-cloning know-how to create an artificial model of a consumer’s voice. The function was designed for people who find themselves prone to dropping their skill to talk, similar to these with a current prognosis of amyotrophic lateral sclerosis or one other degenerative illness, and offers them a option to protect their talking voice in order that their associates, family members and family members can hear from them lengthy into the longer term.
I’ll throw in yet another promising accessibility breakthrough: A analysis crew on the College of Texas at Austin announced this 12 months that it had used A.I. to develop a “noninvasive language decoder” that may translate ideas into speech — learn individuals’s minds, basically. This sort of know-how, which makes use of an A.I. language mannequin to decode mind exercise from fMRI scans, seems like science fiction. However it may make it simpler for individuals with speech loss or paralysis to speak. And it doesn’t require placing an A.I. chip in your mind, which is an added bonus.
To Vertex Prescription drugs and CRISPR Therapeutics, for placing gene modifying to good use
When CRISPR, the Nobel Prize-winning gene modifying software, broke into public consciousness a decade in the past, doomsayers predicted that it’d result in a dystopian world of gene-edited “designer infants” and nightmare eugenics experiments. As a substitute, the know-how has been permitting scientists to make regular progress towards treating a variety of harrowing ailments.
In December, the Food and Drug Administration approved the primary gene-editing remedy for people — a therapy for sickle cell illness, referred to as Exa-cel, that was collectively developed by Vertex Prescription drugs of Boston and CRISPR Therapeutics of Switzerland.
Exa-cel makes use of CRISPR to edit the gene accountable for sickle cell, a debilitating blood illness that impacts roughly 100,000 People, most of whom are Black. Whereas it’s nonetheless wildly costly and troublesome to manage, the therapy affords new hope to sickle cell sufferers who’ve entry to it.
To Brent Seales, Nat Friedman and Daniel Gross, for utilizing A.I. to unlock antiquity’s secrets and techniques
One of the vital enjoyable interviews I did on my podcast this year was with Brent Seales, a professor on the College of Kentucky who has spent the previous 20 years making an attempt to decipher a set of historical papyrus manuscripts referred to as the Herculaneum Scrolls. The scrolls, which belonged to a library owned by Julius Caesar’s father-in-law, had been buried beneath a mountain of ash in 79 A.D. through the eruption of Mount Vesuvius. They had been so totally carbonized that they couldn’t be opened with out ruining them.
Now, A.I. has made it attainable to learn these scrolls with out opening them. And this 12 months, Dr. Seales teamed up with two tech buyers, Nat Friedman and Daniel Gross, to launch the Vesuvius Challenge — providing prizes of as much as $1 million to anybody who efficiently deciphers the scrolls.
The grand prize has nonetheless not been gained. However the competitors sparked a frenzy of curiosity from novice historical past buffs, and this 12 months a 21-year-old laptop science pupil, Luke Farritor, won a $40,000 intermediate prize for deciphering a single phrase — “purple” — from one of many scrolls. I like the concept of utilizing A.I. to unlock knowledge from the traditional previous, and I like the public-minded spirit of this competitors.
To Waymo, for taking the sluggish street to self-driving
I spent quite a lot of time in 2023 being shuttled round San Francisco in self-driving automobiles. Robotic taxis are a controversial know-how — and there are nonetheless loads of kinks to be labored out — however for probably the most half I purchase the concept self-driving automobiles will finally make our roads safer by changing fallible, distracted human drivers with always-alert A.I. chauffeurs.
Cruise, one of many two corporations that had been giving robotic taxi rides in San Francisco, has imploded in recent days, after one in all its autos struck and dragged a lady who had been hit by one other automotive. California regulators stated the corporate had misled them concerning the incident; Cruise pulled its automobiles from the streets, and its chief govt, Kyle Vogt, stepped down.
However not all self-driving automobiles are created equal, and this 12 months I used to be grateful for the comparatively sluggish, methodical method taken by Cruise’s competitor, Waymo.
Waymo, which was spun out of Google in 2016, has been logging miles on public roads for greater than a decade, and it reveals. The half-dozen rides I took in Waymo automobiles this 12 months felt safer and smoother than the Cruise rides I took. And Waymo’s security knowledge is compelling: Based on a study the company conducted with Swiss Re, an insurance coverage agency, in 3.8 million self-driving miles Waymo’s automobiles had been considerably much less more likely to trigger property harm than human-driven automobiles, and led to no bodily damage claims in any respect.
I’ll put my playing cards on the desk: I like self-driving automobiles, and I feel society might be higher off as soon as they’re widespread. However they need to be secure, and Waymo’s slow-and-steady method appears higher suited to the duty.
To the Nationwide Institute of Requirements and Know-how, for managing America’s A.I. transition
One of many extra stunning — and, to my thoughts, heartening — tech developments of 2023 was seeing governments around the globe become involved in making an attempt to know and regulate A.I.
However all that involvement requires work — and in the US, quite a lot of that work has fallen to the Nationwide Institute of Requirements and Know-how, a small federal company that was beforehand higher recognized for issues like ensuring clocks and scales had been correctly calibrated.
The Biden administration’s govt order on synthetic intelligence, launched in October, designated NIST as one of many main federal companies accountable for maintaining tabs on A.I. progress and mitigating its dangers. The order directs the agency to develop methods of testing A.I. programs for security, provide you with workouts to assist A.I. corporations determine doubtlessly dangerous makes use of of their merchandise, and produce analysis and pointers for watermarking A.I.-generated content material, amongst different issues.
NIST, which employs about 3,400 individuals and has an annual finances of $1.24 billion, is tiny in contrast with different federal companies doing vital security work. (For scale: The Division of Homeland Safety has an annual finances of almost $100 billion.) However it’s vital that the federal government construct up its personal A.I. capabilities to successfully regulate the advances being made by private-sector A.I. labs, and we’ll want to speculate extra within the work being finished by NIST and different companies with the intention to give ourselves a preventing probability.
And on that notice: Blissful holidays, and see you subsequent 12 months!