Sam 在与韩国硅谷特派记者团的会面中罕见的透露了比较多关于 GPT-5 的信息。 下面是这篇韩文报道中的一些关键信息和全文:
◆OpenAI CEO Sam Altman警告不要低估GPT-5性能改进的幅度,可能会超出预期。

◆每当GPT的下一个模型推出,企业界和日常生活的多个领域都会发生变革甚至消失,需要更多新思维。
◆这是Altman首次如此自信地对外发出关于GPT-5性能的”信号”。
◆Altman和OpenAI的目标是建立”通用人工智能(AGI)”,如果投入大量计算资源加速AGI到来,AI计算所需能源短缺等问题可以轻松解决。
◆Altman表示GPT-5将在高级推理功能上取得重大进步,只要有足够计算资源,就能建立超越人类的AGI。
◆他警告低估GPT-5的改进幅度并以此为基础开展业务将是一个重大错误,GPT-5的改进幅度远超想象。
◆Altman似乎除了”建立AGI”之外对其他事情不太感兴趣,完全专注于AI。
◆他最近致力于全球AI基础设施创新,认为计算将成为未来最重要的货币,非常关注如何以低成本构建大量计算资源。
◆他乐观认为,如果突破计算能力限制实现AGI,AI运行所需的能源短缺等将成为次要问题。通过AGI进行的科学研究将是可持续经济增长的决定性因素。
◆他担忧数据量跟不上AI开发速度,数据短缺可能阻碍AI进步,需要一个能用更少数据学习更多东西的模型。
现场照片

采访全文:
OpenAI的首席执行官(CEO)萨姆·奥特曼(Sam Altman)警告称,不应低估GPT-5性能改进的幅度,这可能会超出预期。每当GPT的下一个模型被开发出来时,无论是企业界还是日常生活的多个领域都必然会发生变革甚至消失,因此他强调需要更多的新思维。这是奥特曼首次如此自信地对外发出关于GPT-5性能的“信号”。他明确表示,他和OpenAI的目标是建立“通用人工智能(AGI)”,并乐观地指出,如果大量计算资源被投入以加速AGI的到来,人工智能(AI)计算所需的能源短缺等现实问题可以轻松得到解决。 14日(当地时间),在与韩国硅谷特派记者团的会面中,奥特曼表示:“我不确定GPT-5何时发布,但它将作为一个在高级推理功能上更进一步的模型,取得重大进步。” 他补充说:“虽然有许多人质疑GPT模型是否有限制,但我可以自信地说‘没有’。” 这体现了他的信心,即只要有足够的计算资源,就能够建立超越人类的AGI。 奥特曼还指出,低估GPT-5的改进幅度并以此为基础开展业务将是一个重大错误。这表明正在开发中的GPT-5的改进幅度远超过想象。他说:“许多初创公司认为,GPT-5的进步只会是小幅度的,因此他们假设进展会缓慢,但我认为这是一个严重错误。” 他强调:“在技术剧变的情况下,像往常一样,他们将被下一代模型‘压倒(steamrolled)’。” 奥特曼似乎对“建立AGI”之外的事情不太感兴趣。据说他对区块链、生物技术等AI以外的其他技术也失去了兴趣。他表示:“过去,我对世界上发生的一切都有一个非常广泛的视角,并能看到我以前从狭窄的角度看不到的东西。” 但他也说:“遗憾的是,现在我完全专注于AI(AI all of the time at full tilt),很难有其他的视角。” 最近,奥特曼一直在致力于全球AI基础设施的创新。在这个过程中,甚至出现了关于“7万亿美元融资”的传言,成为了一个热点话题。奥特曼说:“除了考虑下一代AI模型外,我最近花费最多时间的领域是‘计算(compute)建设’,我越来越确信,计算将成为未来最重要的货币。” 他还说:“但是,世界并没有为足够的计算做好准备,也没有正视这个问题,因此我非常关注如何以尽可能低的成本构建大量的计算资源。” 这意味着他非常关注为实现AGI而获取计算资源的问题。 他还表达了一种乐观观点,认为如果能突破计算能力的限制并实现AGI,那么AI运行所需的能源短缺等将成为“次要问题”。奥特曼说:“我最期待AGI的是,它能加速科学发现,从而更快实现核聚变发电,快速找到解决能源问题的方法。” 他强调:“通过AGI进行的科学研究将是可持续经济增长的几乎唯一动力和决定性因素。” 他还提出了对数据量跟不上AI开发速度的担忧,数据短缺可能会阻碍AI的进步。他说:“从长远来看,人类产生的数据可能会变得不足。” 他补充道:“因此,我们需要一个能够用更少的数据学习更多东西的模型。”

Sam Altman: GPT-5 will be a technological upheaval. Startups are underestimating future models and they will get steamrolled. - He is confident that with enough compute, we will create AGI that surpasses humans (within 5 years, he said recently) - Actually, he’s obsessed with compute. He is confident compute will be the most important currency in the future. - He’s lost interest in technologies other than AGI (like blockchain and biotechnology) - He said he didn’t know when GPT-5 will be released NOTE: All this is from a translation of a Korean language article, so take this all with a grain of salt. If you speak fluent Korean please help with any important mistranslations! “CEO Altman, who met with Korean Silicon Valley correspondents on the 14th, said, “I don’t know when GPT-5 will be released, but it will make great progress as a model that takes a leap forward in advanced inference functions.” He added, “What are the limitations of GPT? “There are many questions as to whether it exists, but I will confidently say ‘no.’” We are confident that there are no limits to the GPT model and that if sufficient computational resources are invested, it will not be difficult to build AGI that surpasses humans. “I am confident that computing will become the most important currency in the future.” He stressed, “The world is underprepared for the rapidly increasing demand for computing,” and added, “We need to think about what it takes to build massive amounts of computing as cheaply as possible.” CEO Altman also opined that underestimating the improvement margin of the developing GPT-5 and deploying business accordingly would be a big mistake. This implies that the improvement margin of GPT-5 is beyond imagination. He mentioned, “Many startups are happy assuming that GPT-5 will only make slight progress rather than significant advancements (since it presents more business opportunities), but I think this is a big mistake. In this case, as often happens when technological upheavals occur, they will be ‘steamrolled’ by the next-generation model.” Altman appeared to have no interest other than in ‘building AGI’. His interest seems to have faded in other technologies, including blockchain and biotechnology, beyond AI. He said, “In the past, I had a broad perspective on everything happening in the world and could see things I couldn’t from a narrow perspective. Unfortunately, these days, I am entirely focused on AI (AI all of the time at full tilt), making it difficult to have other perspectives.”