Sam Shead

Really good piece this on the lurking challenges for the technology that everyone’s talking about but many people don’t know what to do with. There’s about to be a huge bloodbath for VC-backed lightweight LLM wrappers. But also big questions for OpenAI’s business models. An Enterprise model for ChatGPT might be a thing, but that interface isn’t great for everyone and a lot of the value already freely exists in Bing.

42

1 Comment

Nearly half the world's population goes to the polls this year, including us in the UK. It's also the first big election season - Indian, Mexico, US, UK, EU - since generative AI tools have been really widely available. After years in which mis and disinformation has disrupted and reshaped our democracies, the stakes are now even higher. The team at Demos have been thinking about what could practically be done to help protect democracy. It's too late in the UK for meaningful legislation (the Online Safety Act took five years and ended up a shadow of what it originally promised), the tech companies are working on watermarking and tracking AI generated content but the technical solutions aren't good enough yet. What's left for immediate action? We're calling on the UK political parties to come together to agree how they will use generative AI at the election in order to model best practice and start to shape the norms of what's acceptable use. This is me talking about it on the radio yesterday with Jimmy Wales, founder of Wikipedia, who, along with a wide range of experts and leading voices are backing our call. Working with the brilliant Full Fact we have drafted an oven-ready agreement the parties should sign up to committing to: • Not using generative AI to create deepfakes designed to mislead • Labelling where they are using generative AI to improve transparency and trust • Not amplifying content they suspect could have been created using generative AI • Giving their campaigners guidance on how to use generative AI responsibly.... It won't solve the bigger problem of malicious actors and state interference, but it will start to show world class leadership in this area... Read the full agreement here: https://lnkd.in/eW7YEdeh Read our briefing paper by the brilliant Elizabeth Seger here: https://lnkd.in/eYH7HFzM

50

4 Comments

Kate Jones, Digital Regulation Cooperation Forum (DRCF) CEO: "AI is not a sector, it's going to permeate everywhere" The UK has avoided new, over-arching regulation on AI. But AI companies still need to navigate regulation: enterprise customers can hesitate due to risks and copyright/IP questions shape product and technical choices, while 13 major regulators have set out their 'strategic approaches to AI. In that context, the DRCF has launched a new 'AI & Digital Hub': a pilot service support startups and innovators with questions about the AI regulatory frontier. So we spoke to Kate Jones, CEO, to hear more about her journey, the role of the DRCF in AI regulation, and what startups should expect from the new initiative: https://lnkd.in/gGSPQkA6

19

Old-fashioned rule-based NLP can still generate far more clever jokes than LLMs, in my opinion. But you be the judge! Which "champagne for my real friends, real pain for my sham friends"-style toast is better? 1. ChatGPT's "Confetti for my loyal friends, bitter truth for my deceitful friends" 2. My algorithm's: "focus for my real friends, real cuss for my faux friends" 3. For the #datascience practitioners out there, my algorithm's "metrics for my real friends, real tricks for my meh friends" Read more about how I generated these -- using WordNet, the CMU Pronouncing Dictionary, GLoVe and #python -- on my blog: https://lnkd.in/e5kRduaw

2

In the latest edition of The Prompt: wild speculation on the mysterious gpt2-chatbot, the future of expressive and realistic avatars and the challenge of developing AI coding assistants. https://lnkd.in/gpG8-RAR

24

3 Comments

What's been buzzing the loudest for me so far at HLTH Europe? AI, AI, and more AI, of course. Tongues have been tripping over those two letters in every conversation I've had today or listened to including these nuggets: Michael Chen of Nuclera: "Generative AI is wonderful because it brings more hypothesis-making power for the drug discovery researcher. But ultimately, we all know that AI isn't enough. You need to have the power at the wet lab benchtop in order to validate experiments and bring hits and needs into the local trials, and the quality of these hits and leads that ultimately will help the success rates in phase one through three clinical trials, which often don't rear their ugly heads until it's too late." Geraldine O'Keeffe of EQT Group: "There's a huge amount of investment opportunity in terms of AI-, SaaS-based models, in big data and machine learning in diagnostics, precision medicine...delivering the care you need better, faster, cheaper." Karen DeSalvo of Google: "The availability of AI, the ubiquity of it, and just how quickly that is moving [has been significant in the evolution of AI in healthcare]. What I'm hoping now is the next generation of this technology allows us to democratise access to health for people all over the world, and that we can start to do that much more quickly and much more equitably." Chiara Bucciarelli-Ducci, MD, PhD and Marlies Schijven on clinicians' take on AI: "To scale up adoption of AI in healthcare, industry needs to demonstrate to clinicians and healthcare providers the cost effectiveness and cost efficiencies that the technology can bring at every level." "More often than not, we find that AI does work and helps to bring efficiencies, but it also carries risk. So it needs to be properly explained." Agree, disagree, or follow-up comments on these soundbites? #genAI #AIeverywhere #medicalAI #healthtech #digitalhealth #AIfordrugdiscovery #techbio #healthAI #clinicalAI

17

3 Comments

Since ChatGPT and similarly powerful AI systems exploded on the scene less than two years ago, people have been amazed by their stunning ability to mimic human speech — and by their frequently flabbergasting errors. AI researchers and writers have come to call those falsehoods and mistakes "hallucinations." But a group of philosophers in Scotland is calling "b.s." on that term. The group — lecturers at the University of Glasgow — argue that calling such mistakes "hallucinations" gives too much credit to the systems and how they work. Instead, taking a page from a famous two-decade old book, they think it's more appropriate to think of the output of such systems as, well, b.s. I spoke with the group — Michael Townsen Hicks, Joe Slater and Joseph Humphries — as well as with AI experts James Landay, Chirag Shah and Alex Hanna, Ph.D. about whether and why b.s. is an apt metaphor for ChatGPT.

37

9 Comments

#ElonMusk has now completely caged the #Twitter bird Apart from a sole embed-sharing page, the iconic bird (or even the blue colourway) is nowhere to be seen anymore. I reached out to some seasoned users of Twitter, trying to find out if Musk's #X has any charm left at all, nearly two years after the hostile $44bn takover. Some of the responses: "It is not as welcoming as the blue bird was, and heck, it is a pain to find on the phone app menu if you have dark theme on, but one gets used to it." "With the name and bird gone, [X] just looks like some ultra-rich technocrat's fantasy and a former shell of what meant something to the rest of us" and even, "People die and the world moves on. What is an online brand?" Light-heartedness aside, almost everyone admitted how X is now chock-a-block with #spam and #bots and how there has been a visible dip in #traffic if you aren't ready to loosen your purse strings. Most have moved on to #Instagram for some joy, or to #LinkedIn for some better traffic. I had fun writing this one (trashing #Musk is now my new favourite pastime), and I hope you enjoy my latest edition of #DigitalDisconnect: https://bit.ly/3WS4nBn For more interesting long-forms and thought provokers, check out ABP Live

6

At a recent internal event, we had a great conversation about the balance between owned storytelling on our platforms, and earned media. The dialogue started with a simple question -- when should you consider one over the other, and why? Owned offers a controlled environment to craft and refine your narrative. Take our recent quantum partnership story as an example. This is a super complex topic with many difficult-to-grasp concepts. It involves work with a partner, who is equally invested in getting these complicated details right. It is the kind of story that benefits from a longer runway up front to get it right, and here it is presented concisely with supporting assets and a concrete example that allows audiences to learn as they read the story.   When it comes to earned media, the dynamics shift. Here, newsworthiness and external validation play pivotal roles. The credibility and often the reach garnered through earned media often surpasses what we can achieve through owned channels alone. So as we think about news moments and engagement with influencers, we pivot our owned assets to support these efforts, rather than lead them. As is so often the case :) we have to find the right balance, the right tool for the moment. Every organization needs to have the capacity to tell its own story, and to find the right time to lead there. This is what I think of as the "voice" of the company. And, the understanding that not every piece of news should lead with that voice.

198

6 Comments

A day when people can interact directly with computers using their thoughts could be on the horizon. Several companies, including Elon Musk’s Neuralink, have begun preliminary human trials of brain-computer interfaces - devices that decode the electrical signals in their brain and translate them into digital bits. For the latest episode of The Wall Street Journal's Future of Everything podcast, I spoke with neurosurgeon and Precision Neuroscience co-founder Benjamin Rapoport, a company working on brain-computer interfaces. We talked about how the technology works and how these implants could improve the lives of hundreds of thousands of people who could gain the ability to independently engage with the digital world. Take a listen!

11

1 Comment

“AI models are not, of course, people. But as they become more capable, we believe we can—and should—try to train them to behave well.” If you have any interest in AI you should read this. I’m really glad Anthropic put it out. It highlights the kind of real world, immediate decisions that need to be taken by teams generating LLMs, along with the whole range of value judgments and potential impacts they involve. Even just this one question of ‘character’ illuminates the complexity of the philosophical issues.

30

2 Comments

As OpenUK publishes Phase Two of the Open Manifesto Report to guide the incoming UK Government in driving the adoption of open source software, increasing employment, and boosting the UK economy, we have three reports from George Lawton with highlights and analysis. First article here: bit.ly/4cxUz4u

4

A lot is being written about AI, stories painting the tech as either revolutionary or pure evil, as either black or white. I find Madhumita Murgia's Code Dependent as one of the most important reads on AI, especially for the so-called Global South. Here she explores the gray, murkier areas of the AI revolution, putting the common folk and their stories into the spotlight. The book tells the story of data labellers of Kenya, on whose backs the algorithms of self-driving cars are trained, and of a mother from the Netherlands who one day finds her child added to a list of potential criminals, a list created not by humans but a machine with all its racial bias. She tells the story of thousands of gig workers working for delivery apps, often underpaid, but also nobody to turn to to register their grievances, because, well, a machine is their boss. Also, what is glaring is the global North-South divide in AI development, where the bigwigs of Silicon Valley make hefty sums while the mundane-repetitive task of training and deploying the tech using real-world data is outsourced to the poor populace of the developing world. Murgia writes about “data colonialism” in more places than one, and reading through the book, one cannot help but draw the parallels between the likes of the East India Company and the Big Techs of today. #artificialtechnology #ai #technology #codedependent #books #reading

11

1 Comment

As a tech journalist, it's not every day you get to write about the Pope, but here we are! 😊 Pope Francis has recently been very outspoken about AI regulations and a need for world leaders to harness the technology for the 'common good'. He is expected to speak about the technology at the G7 Summit in June 2024. Read more below. #AI #AIethics #PopeFrancis #Technology #EUAIAct #Europe #G7Summit

15

One of the best parts of my job working at Cipher News is learning from my esteemed colleagues. You should learn from them, too!!! A few highlights this week: Anca Gurzu on how the EU nations need to work together in a more comprehensive manner to be competitive in the clean tech race: https://lnkd.in/eXPZsBbN Amena H. Saiyid has a look at where critical minerals are refined. China dominates. Nickel is the only critical mineral where China does not COMPLETELY dominate. Indonesia holds a plurality of the nickel refining at 37%, but even there, 28% of the nickel refining happens in China: https://lnkd.in/eFja-cFE (Chart shown below) And Bill Spindle highlighted this fantastic The New York Times story https://lnkd.in/e423tcnG in one of his always incisive hot takes on the news of the day. Says Bill: "However problematic Saudi oil and gas remain for the world, the Kingdom has the resources — money, sun and wind — to move the needle globally on renewable energy and even emissions." https://lnkd.in/ePftiUEQ And our always impressive fearless leader Amy Harder called out this Bloomberg News piece: https://lnkd.in/eqjYCfrb about the carbon offsets controversy within The Science Based Targets initiative. "This controversy is emblematic of society’s broader debate of how best to tackle climate change and what role offsets and corporations should play," says Amy. https://lnkd.in/euKSznjT Make sure you are reading CipherNews.com each day to get smart on clean tech and the energy transition.

18

1 Comment

Everyone in Silicon Valley loves open source-software. Much of the tech economy is built on applications that are essentially free to use, customize and redistribute — think Linux, Android and the web browsing engines underneath Chrome and Safari. But there's a growing debate ranging from Northern California to Washington, D.C., to well beyond over whether the open-source ethos should be applied to artificial intelligence. Some in national security and public policy circles are raising concerns that making powerful AI models freely available, the U.S. will be giving a leg up to adversaries such as China or making it easier for malicious actors to use the technology for harmful ends. Some technologists, on the other hand, argue that such worries are overblown and limiting the open-sourcing of AI will inhibit development of the technology. However, unlike similar technology debates in the past, this one doesn't divide cleanly, between technologists and national security or policy advocates. There are prominent folks here in Silicon Valley that are arguing against open-source AI. Here's a look at the dispute:

9

Is a reckoning coming for venture firms? Plenty of VC firms have jumped on the AI bandwagon, looking for startups to back. But the tech could ultimately lead to a serious reduction in headcount for VCs, writes Ben Bergman. But AI isn't the only risk for the industry. According to one VC, the space has simply grown too big and is going to see a massive contraction, writes Blake Dodge. VCs, who often tout their ability to help solve big problems, might have their biggest task yet: Saving themselves. Read more Business Insider #venturecapital #vc #investing #ai https://lnkd.in/g9B83Dq8

19

Weekend reading: The Artificial Intelligence Revolution https://lnkd.in/dhh2Zf2C This is a brilliant and funny article about where we could be going with AI. The key thing is that it’s from 2015, so it’s interesting to see how much of what we see now was predicted nine years ago. It’s also a rare article that is insightful, inspiring, and scary.   Selected quote: “Superintelligence of that magnitude is not something we can remotely grasp, any more than a bumblebee can wrap its head around Keynesian Economics. In our world, smart means a 130 IQ, and stupid means an 85 IQ - we don’t have a word for an IQ of 12,952.”   This is why you should always add “please” when making requests on ChatGPT – if there is an AI Robot uprising, I’m hoping those of us who were kind and polite will be spared.

4