 |
|
|
|||||
Curious Kids is a series for children of all ages. If you have a question youd like an expert to answer, send it to CuriousKidsUS@theconversation.com. Is the whole universe just a simulation? Moumita B., age 13, Dhaka, Bangladesh How do you know anything is real? Some things you can see directly, like your fingers. Other things, like your chin, you need a mirror or a camera to see. Other things cant be seen, but you believe in them because a parent or a teacher told you, or you read it in a book. As a physicist, I use sensitive scientific instruments and complicated math to try to figure out whats real and whats not. But none of these sources of information is entirely reliable: Scientific measurements can be wrong, my calculations can have errors, and even your eyes can deceive you, like the dress that broke the internet because nobody could agree on what colors it was. Because every source of informationeven your teacherscan trick you some of the time, some people have always wondered whether we can ever trust any information. If you cant trust anything, are you sure youre awake? Thousands of years ago, Chinese philosopher Zhuangzi dreamed he was a butterfly and realized that he might actually be a butterfly dreaming he was a human. Plato wondered whether all we see could just be shadows of true objects. Maybe the world we live in our whole lives inside isnt the real one; maybe its more like a big video game, or the movie The Matrix. The simulation hypothesis The simulation hypothesis is a modern attempt to use logic and observations about technology to finally answer these questions and prove that were probably living in something like a giant video game. Twenty years ago, a philosopher named Nick Bostrom made such an argument, based on the fact that video games, virtual reality, and artificial intelligence were improving rapidly. That trend has continued, so that today people can jump into immersive virtual reality or talk to seemingly conscious artificial beings. Bostrom projected these technological trends into the future and imagined a world in which wed be able to realistically simulate trillions of human beings. He also suggested that if someone could create a simulation of you that seemed just like you from the outside, it would feel just like you inside, with all of your thoughts and feelings. Suppose thats right. Suppose that sometime in, say, the 31st century, humanity will be able to simulate whatever they want. Some of them will probably be fans of the 21st century and will run many different simulations of our world so that they can learn about us, or just be amused. Heres Bostroms shocking logical argument: If the 21st century planet Earth only ever existed one time, but it will eventually get simulated trillions of times, and if the simulations are so good that the people in the simulation feel just like real people, then youre probably living on one of the trillions of simulations of the Earth, not on the one original Earth. This argument would be even more convincing if you actually could run powerful simulations today; but as long as you believe that people will run those simulations someday, then you logically should believe that youre probably living in one today. Scientist Neil deGrasse Tyson explains the simulation hypothesis and why he thinks the odds are about 50-50 were part of a virtual reality. Signs were living in a simulation . . . or not If we are living in a simulation, does that explain anything? Maybe the simulation has glitches, and thats why your phone wasnt where you were sure you left it, or how you knew something was going to happen before it did, or why that dress on the internet looked so weird. There are more fundamental ways in which our world resembles a simulation. There is a particular length, much smaller than an atom, beyond which physicists theories about the universe break down. And we cant see anything more than about 50 billion light-years away because the light hasnt had time to reach us since the Big Bang. That sounds suspiciously like a computer game where you cant see anything smaller than a pixel or anything beyond the edge of the screen. Of course, there are other explanations for all of that stuff. Lets face it: You might have misremembered where you put your phone. But Bostroms argument doesnt require any scientific proof. Its logically true as long as you really believe that many powerful simulations will exist in the future. Thats why famous scientists like Neil deGrasse Tyson and tech titans like Elon Musk have been convinced of it, though Tyson now puts the odds at 50-50. Others of us are more skeptical. The technology required to run such large and realistic simulations is so powerful that Bostrom describes such simulators as godlike, and he admits that humanity may never get that good at simulations. Even though it is far from being resolved, the simulation hypothesis is an impressive logical and philosophical argument that has challenged our fundamental notions of reality and captured the imaginations of millions. Hello, curious kids! Do you have a question youd like an expert to answer? Ask an adult to send your question to CuriousKidsUS@theconversation.com. Please tell us your name, age, and the city where you live. And since curiosity has no age limitadults, let us know what youre wondering, too. We wont be able to answer every question, but we will do our best. Zeb Rocklin is an associate professor of physics at Georgia Institute of Technology. This article is republished from The Conversation under a Creative Commons license. Read the original article.
Category:
E-Commerce
A few months ago, I walked into the office of one of our customers, a publicly traded vertical software company with tens of thousands of small business customers. I expected to meet a traditional support team with rows of agents on the phones, sitting at computers triaging tickets. Instead, it looked more like a control room. There were specialists monitoring dashboards, tuning AI behavior, debugging API failures, and iterating on knowledge workflows. One team member who had started their career handling customer questions over chat and email (resetting passwords, explaining features, troubleshooting one-off issues, and escalating bugs) was now writing Python scripts to automate routing. Another was building quality-scoring models for the companys AI agent. This seemed markedly different from the hyperbole Id been hearing about customer support roles going away in large part due to AI. What I was seeing across our customer base looked more like a shift in how support work is defined. So I decided to take a closer look. I analyzed 21 customer support job postings across AI-native companies, high-growth startups, and enterprise SaaS. These jobs run the gamut from technical support for complex software products to more transactional, commercial support involving billing and other common issues. What I found was that customer support is being rebuilt around AI-native workflows and systems-level thinking. Yes, responding to individual tickets is still important, but roles are designing and operating the technical systems that resolve customer issues at scale. The result is a new kind of support role, one thats part operator, part technologist, part strategist. AI Skills Are Now Table Stakes For most of the last two decades, support hiring optimized for communication skills and product familiarity. But that baseline is now gone. Across the 21 job postings I analyzed, nearly three-quarters explicitly required experience with AI tools, automation platforms, or conversational AI systems. These roles are about configuring, monitoring, and improving the AI systems over time. They are reviewing conversation logs, auditing AI behavior, and identifying failure modes. In other words, AI literacy has become the baseline for modern support work. If you dont understand how AI systems behave, you cant support the customers relying on them. More than half of the roles I analyzed required candidates to debug APIs, analyze logs, write SQL queries, or script automations in Python or Bash. Many expected familiarity with cloud infrastructure, observability tools, or version control systems like Git. That would have been unthinkable in support job descriptions even five years ago. But it makes sense. When AI systems fail, they fail at scale. Diagnosing those failures requires technical fluency like understanding how models interact with external systems and when an issue is rooted in configuration versus product logic. The job has evolved from fixing problems ticket by ticket to preventing the next thousand tickets. Humans are Needed to Solve Harder Problems Once AI becomes part of the support workflow, the nature of the work becomes more technical. One support leader I spoke with at a company that now contains more than 80% of its tickets with AI put it plainly: once automation handles the easy questions, the work left behind gets harder. The same frontline agents who used to focus on quick wins are now handling the most frustrated customers and edge cases, and theyve had to scale up their skills accordingly. In practice, this often looks like a customer trying to complete a critical workflow, like syncing data between systems before running billing. An AI agent starts by working off documentation that a subject matter expert has synthesized from multiple functions across the company. From there, the AI agent can confirm that everything is configured correctly. However, the AI agent may not be integrated to the right underlying system that failed silently hours earlier. The customer follows the guidance, only to discover downstream that data didnt move as expected. When the issue escalates, the subject matter expert has to reconstruct what happened across systems, reason through what the AI agent missed, and help the customer recover without losing trust. This is the kind of end-to-end work that AI still cant do on its own. It requires both technical fluency to trace failures across disparate systems, in addition to human judgement to decide what can be fixed immediately versus what needs deeper product or engineering intervention. In this way, support has become less about answering questions out of the manual, and more about creating the manual and solving the problems that it doesnt cover. The Hybrid HumanAI Model Is the Default Despite widespread fear about AI replacing support jobs, not a single posting I analyzed suggested that support would be 100% automated in the future. Instead, nearly every role gravitated toward a hybrid model where AI handles routine interactions, while humans oversee quality and continuously improve the system. This makes sense when you consider the fact that 95% of customer support leaders said they would retain human agents in their operations to help define AIs role when surveyed by Gartner last year. Titles like AI Support Specialist, AI Quality Analyst, and Support Operations Specialist were almost entirely focused on orchestration, designing escalation logic and defining when humans step in. This is where the earlier control room image becomes reality. The work of humans changes from simply answering questions to actually shaping systems. Taken together, these trends point to a single conclusion: customer support is specializing. The repetitive work is going away, but the judgment-heavy, technical work is expanding. That shift is already visible in how companies hire. The question now becomes whether organizations (and workers) are ready to adapt fast enough.
Category:
E-Commerce
Forget about the big game on Sunday. Two heavyweights have been battling it out this week over a topic thats become all-too-familiar over the years: advertising creep. Its a tale as old as time, in some respects. Many a CEO have proudly declared that their companys platform or services will remain ad-free, only to later succumb to the lure of all that advertising revenue and embrace it. And thats creating a new divide among AI platformsone that will play out to the worlds largest TV viewing audience during the Super Bowl. Among the nearly dozen AI-related ads on Sunday will be two 60-second spots each for OpenAI and Anthropic. While OpenAI will use its time to tout how its ChatGPT helps people build things with a real-world impact, Anthropic is taking a swipe at its much-larger competitors recent decision to incorporate ads into the platformand its virtuous decision to keep its Claude AI assistant ad-free. This off-field action has already made for some entertaining viewing. Anthropic dropped four ads this past week set to Dr. Dres 2001 song Whats the Difference, which all conclude with the same warning: Ads are coming to AI. But not Claude. That elicited a lengthy post on X by Sam Altman, OpenAIs CEO, thats since become the topic of much ridicule. While Anthropics Black Mirror-esque ads are extremely clever, theyre also strategic, as the company seeks to position itself much like Apple has with its privacy-first messaging, says John Battelle, an entrepreneur and author. Claudes marketshare may be a tiny fraction of ChatGPTs, but Anthropics swipe at OpenAIs reversal on advertising was well-playedand it makes an important point, according to Battelle. I applaud what they did,” he tells Fast Company. “This needs to be talked about.” Pivot to revenue For Altman, the advertising U-turn took about 15 months. In October 2024, he said that he viewed advertising as a last resort as a business model for OpenAI, only to seemingly reverse course a few weeks ago. Hes hardly the first to do soand for good reason. Theres no question that the only business model that you can employ if you want to scale a business is advertising, Battelle says. Advertising . . . finds a way Recent decades have seen a well-trod path for companies that transitioned to an advertising-supported business model from one that was originally based on subscriptions or, seemingly, free for users. Advertising has slowlyand sometimes not so slowlycrept into almost every major platform controlled by a large tech company, whether it’s social networks like Facebook or streaming services like Netflix, to name a few. But Google perhaps offers the canonical example of a companys founders finding their advertising religion and walking back prior vows to remain ad-free, says Battelle, whose book The Search: How Google and Its Rivals Rewrote the Rules of Business and Transformed Our Culture” was published 20 years ago by Penguin In a 1998 white paper laying out their vision for Google, cofounders Sergey Brin and Larry Page didnt mince words about the potential mixed incentives of a commercial search engine with a business model predicated on advertising. The goals of the advertising business model do not always correspond to providing quality search to users, wrote the duo, then-PhD students at Stanford University. We expect that advertising funded search engines will be inherently biased towards the advertisers and away from the needs of the consumers. In 2000, Google began selling advertisements. To say it’s become a big business would be an understatement. Parent company Alphabet reported this past week that Google advertising accounted for more than 72% of its $113.8 billion revenue in the fourth quarter. A hill to die on For some founders, advertising has been a hill theyve been willing to die on. Clashes about advertising on WhatsApp were reportedly among a variety of reasons why cofounder Jan Koum ultimately left parent company Facebook (now Meta Platforms) in 2018, four years after the messaging app was acquired. Koum had long expressed his disdain for advertising and how it worsened a product. He vowed to keep advertising out of an app where people communicate with friends and family, and his first-ever tweet was a quote from the movie Fight Club that pulls no punches: Advertising has us chasing cars and clothes, working jobs we hate so we can buy shit we dont need. Last year, WhatsApp debuted ads in the app. The future of advertising on AI Like many before them, the founders of AI companies must now determine where their line in the sand is on advertising. And its no surprise that advertising has eventually found its way into AI platforms, as Battelle laid out in a recent blog post. As Google, Facebook, Uber, and other companies have taught us, advertising creep does gradually change the services we know and love, Battelle says. In some cases, like with Instagram, the product today barely resembles its pre-advertising self. When you commit to advertising, you commit to the incentives of advertising, he says. Even if advertising on AI platforms will inevitably evolve with time, Battelle says an imperative question now is what that future will look likeand particularly if social media is the reference point. Man, things are not going to be pretty if the business model that drove social media also drives generative AI.
Category:
E-Commerce
All news |
||||||||||||||||||
|
||||||||||||||||||