TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

If AI seems smarter, it's thanks to smarter human trainers

182 点作者 getwiththeprog8 个月前

16 条评论

throwaway_29688 个月前
Throwaway account here. I recently spent a few months as a trainer for a major AI company&#x27;s project. The well-paid gig mainly involved crafting specialized, reasoning-heavy questions that were supposed to stump the current top models. Most of the trainers had PhDs, and the company&#x27;s idea was to use our questions to benchmark future AI systems.<p>It was a real challenge. I managed to come up with a handful of questions that tripped up the models, but it was clear they stumbled for pretty mundane reasons—outdated info or faulty string parsing due to tokenization. A common gripe among the trainers was the project&#x27;s insistence on questions with clear-cut right&#x2F;wrong answers. Many of us worked in fields where good research tends to be more nuanced and open to interpretation. I saw plenty of questions from other trainers that only had definitive answers if you bought into specific (and often contentious) theoretical frameworks in psychology, sociology, linguistics, history, and so on.<p>The AI company people running the projects seemed a bit out of their depth, too. Their detailed guidelines for us actually contained some fundamental contradictions that they had missed. (Ironically, when I ran those guidelines by Claude, ChatGPT, and Gemini, they all spotted the issues straight away.)<p>After finishing the project, I came away even more impressed by how smart the current models can be.
评论 #41685949 未加载
评论 #41685159 未加载
评论 #41684093 未加载
评论 #41688756 未加载
评论 #41684420 未加载
评论 #41684242 未加载
JCharante8 个月前
&gt; AI models now require trainers with advanced degrees<p>Companies that create data for FM (foundational model) companies have been hiring people with degrees for years<p>&gt; Invisible Tech employs 5,000 specialized trainers globally<p>Some of those companies have almost a million freelancers on their platforms, so 5k is honestly kinda medium sized.<p>&gt; It takes smart humans to avoid hallucinations in AI<p>Many smart humans fail at critical thinking. I&#x27;ve seen people with masters fail at spotting hallucinations in elementary level word problems.
评论 #41680567 未加载
评论 #41683186 未加载
评论 #41681140 未加载
Stem00378 个月前
AI, at least in its current form, is not so much replacing human expertise as it is augmenting and redistributing it.
评论 #41680814 未加载
评论 #41683232 未加载
recursive8 个月前
It kind of seems like it got dumber to me. Maybe because my first exposure to it was so magical. But now, I just notice all the ways it&#x27;s wrong.
评论 #41683992 未加载
theptip8 个月前
I feel this is one of the major ways that most pundits failed with their “the data is going to run out” predictions.<p>First and foremost a chatbot generates plenty of new data (plus feedback!), but you can also commission new high-quality content.<p>Karpathy recently commented that GPT-3 needs so many parameters because most of the training set is garbage, and that he expects eventually a GPT-2 sized model could reach GPT-3 level, if trained exclusively on high-quality textbooks.<p>This is one of the ways you get textbooks to push the frontier capabilities.
评论 #41683558 未加载
评论 #41682073 未加载
CamperBob28 个月前
Which is fine. If all AI does is represent human knowledge in a way that makes it explainable and transformable rather than merely searchable, then the hype is justified... along with Google&#x27;s howling, terrified panic.<p>The role played by humans on the training side is of little interest when considering the technology from a user&#x27;s perspective.
评论 #41681046 未加载
评论 #41680371 未加载
yawnxyz8 个月前
&quot;raw dogging&quot; non-RLHF&#x27;d language models (and getting good and unique output) is going to be a rare and sought-after skill soon. It&#x27;s going to be a new art form<p>someone should write a story about that!
评论 #41683813 未加载
SamGyamfi8 个月前
There is a cost-quality tradeoff companies are willing to make for AI model training using synthetic data. It shows up fairly often with AI research labs and their papers. There are also upcoming tools that remove the noise that would trip up some advanced models during annotation. Knowing this, I don&#x27;t think the &quot;human-labeled data is better&quot; argument will last that long.
GaggiX8 个月前
Let&#x27;s not ignore better architectures, training techniques and computing power.
评论 #41680882 未加载
评论 #41680344 未加载
评论 #41680532 未加载
jwrallie8 个月前
I wonder how much prompt skill is actually influencing the quality of the response.<p>After using LLMs daily for some time, I have developed a feeling on how to phrase my requests as to get better quality answers.<p>For example, ensure that it can process the information linearly, like asking to classify items in a list and adding the label directly after the item so that the order remains, instead of allowing it to create multiple lists as the output (which it tends to do by default).<p>So, at least for me, the prompts are getting smarter.
评论 #41684140 未加载
评论 #41706478 未加载
fsndz8 个月前
&gt; &quot;In the early years, getting AI models like ChatGPT or its rival Cohere to spit out human-like responses required vast teams of low-cost workers helping models distinguish basic facts such as if an image was of a car or a carrot.&quot;<p>Starting the article by comparing what is necessary for ChatGPT to work and image labelling is a bit weird
tiku8 个月前
I was watching the video of someone using openai voice chat, asking for a joke. Then I&#x27;ve tried it myself, asking for a joke and I got the exact same one. (Why don&#x27;t skeletons fight eachother?).<p>Seems like if then else haha.
butz8 个月前
I bet that every &quot;top&quot; performing GenAI has dozens upon dozens of &quot;if&quot; statements to make them seem smart.
bdjsiqoocwk8 个月前
Submarine article placed by Cohere. Wtf is cohere.
wlindley8 个月前
a&#x2F;k&#x2F;a It is all a clever scam. True, or true?
ysofunny8 个月前
I feel weird being stubborn against free tier google gemini<p>I feel as though it &#x27;extracts&#x27; some sort of &quot;smartness&quot; out of me (if any) and then whatever intelligence from me becomes part of google gemini<p>this is why I would never want to pay for using these tools, anything good that comes from me in the chat becomes google&#x27;s by AI training, which is ok so long as it&#x27;s free to use<p>i.e. I won&#x27;t pay to make their stuff better through my own work
评论 #41681075 未加载
评论 #41680626 未加载
评论 #41680597 未加载
评论 #41681396 未加载