Some commentators say investors are paying too much for technology stocks because of misplaced expectations about AI developments. Some commentators say investors are paying too much for technology stocks because of misplaced expectations about AI developments. Some experts have voiced fears a tech meltdown could hit our savings and pensions - here's how to protect yourself T he new year has started as 2025 ended - with share prices booming amid warnings from some that the growth is being driven by overvalued technology stocks. Fears of an "AI bubble" have been voiced by people from the governor of the Bank of England to the head of Google's parent company, Alphabet . Even if you have not actively invested in technology shares, the chances are you have some exposure to companies operating in the sphere.

Large language models (LLMs) have shown promising accuracy in predicting survey responses and policy preferences, which has increased interest in their potential to represent human interests in various domains. Most existing research has focused on "behavioral cloning", effectively evaluating how well models reproduce individuals' expressed preferences. Drawing on theories of political representation, we highlight an underexplored design trade-off: whether AI systems should act as delegates, mirroring expressed preferences, or as trustees, exercising judgment about what best serves an individual's interests. This trade-off is closely related to issues of LLM sycophancy, where models can encourage behavior or validate beliefs that may be aligned with a user's short-term preferences, but is detrimental to their long-term interests. Through a series of experiments simulating votes on various policy issues in the U.S. context, we apply a temporal utility framework that weighs short and long-term interests (simulating a trustee role) and compare voting outcomes to behavior-cloning models (simulating a delegate). We find that trustee-style predictions weighted toward long-term interests produce policy decisions that align more closely with expert consensus on well-understood issues, but also show greater bias toward models' default stances on topics lacking clear agreement. These findings reveal a fundamental trade-off in designing AI systems to represent human interests. Delegate models better preserve user autonomy but may diverge from well-supported policy positions, while trustee models can promote welfare on well-understood issues yet risk paternalism and bias on subjective topics.

Knowledge distillation allows smaller neural networks to emulate the performance of larger, teacher models with reduced computational demands. Traditional methods for Large Language Models (LLMs) often necessitate extensive fine-tuning, which limits their accessibility. To address this, we introduce Trace-of-Thought Prompting, a novel framework designed to distill critical reasoning capabilities from high-resource teacher models (over 8 billion parameters) to low-resource student models (up to 8 billion parameters). This approach leverages problem decomposition to enhance interpretability and facilitate human-in-the-loop interventions. Empirical evaluations on the GSM8K and MATH datasets show that student models achieve accuracy gains of up to 113% on GSM8K and 21% on MATH, with significant improvements particularly notable in smaller models like Llama 2 and Zephyr. Our results suggest a promising pathway for open-source, low-resource models to eventually serve both as both students and teachers, potentially reducing our reliance on high-resource, proprietary models.

This story was produced with support from the Pulitzer Center's AI Accountability Network. Rohtak and New Delhi, India: Dhuli Chand was 102 years old on September 8, 2022, when he led a wedding procession in Rohtak, a district town in the north Indian state of Haryana. As is customary in north Indian weddings, he sat on a chariot in his wedding finery, wearing garlands of Indian rupee notes, while a band played celebratory music and family members and villagers accompanied him. But instead of a bride, Chand was on his way to meet government officials. Chand resorted to the antic to prove to officials that he was not only alive but also lively.

Fox News Flash top headlines are here. Check out what's clicking on Foxnews.com. Mike Berent has spent more than 27 years rushing into burning houses in Detroit, pulling people to safety and ensuring his fellow firefighters get out alive. But as the 52-year-old Detroit Fire Department lieutenant approaches mandatory retirement at age 60, he says one thing is clear: He will need to keep working to make ends meet. "I'm trying to put as much money away as a I can," said Berent, who also works in sales.

Conversational machine reading systems help users answer high-level questions (e.g. determine if they qualify for particular government benefits) when they do not know the exact rules by which the determination is made(e.g. whether they need certain income levels or veteran status). The key challenge is that these rules are only provided in the form of a procedural text (e.g. guidelines from government website) which the system must read to figure out what to ask the user. We present a new conversational machine reading model that jointly extracts a set of decision rules from the procedural text while reasoning about which are entailed by the conversational history and which still need to be edited to create questions for the user. On the recently introduced ShARC conversational machine reading dataset, our Entailment-driven Extract and Edit network (E3) achieves a new state-of-the-art, outperforming existing systems as well as a new BERT-based baseline. In addition, by explicitly highlighting which information still needs to be gathered, E3 provides a more explainable alternative to prior work. We release source code for our models and experiments at https://github.com/vzhong/e3.

We believe that the real key to engaging members at scale is by understanding the context of where they are in their life stage and providing relevant and contextual advice based on what they are hoping to achieve both in the short-term as well as the long-term. Artificial Intelligence can help identify behavioural patterns and provide intelligent and contextual guidance based on the individual's circumstances. This can offer the member a vastly improved customer journey which they are crying out for -- simple, intuitive, engaging and personalised. At a high-level, there are 4 key stages of a member journey of how they engage with their pension and in this series of articles we will tackle how AI can deliver relevant advice to help members save more depending on what stage they find themselves in. The amount of data we create and consume is expanding exponentially.

The big question the blockchain community is only just starting to answer is what will the governance issues look like in the new era? And when most things get to be a smart contract, will they be smart enough? To get an understanding of the future and the big picture, it's helpful to go back to the past when things were barely automated at all. Let's take a look at a worker back in the 70s. In those days, of punch card and cash, you could earn money by working on a car production line, for example.

So what are the "megatrends" in this evolving job market? Tim Baxter, president and COO of Samsung Electronics America, speaks during a press event for CES 2017 at the Mandalay Bay Convention Center on January 4, 2017 in Las Vegas, Nevada. Most of the jobs created in advanced economies don't offer permanent contracts, but involve self-employed or freelance consultants. This means that these people have no social "safety nets" like insurance, medical coverage, social security, or paid vacation. In the U.S., 94% of the new jobs created from 2005 to 2015 fell into this category, giving these workers no protection at all.

Imagine a Sunday morning sometime in the future. As you pour milk on to your cornflakes, it's a reminder that you intended to check out whether the Chinese company Big Dairy Inc is worth a punt. You press a button on your wrist monitor, ResearchBit, and it sends a request to your peer-to-peer investment network for any information on the company. Within minutes investor friends from around the globe are commenting via a private social media network, InvestBook. As you munch through breakfast, you decide to do a bit more digging.