TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Google’s AI thinks I left a Gatorade bottle on the moon

367 pointsby gwintrob7 months ago

35 comments

simonw7 months ago
The linked article describes an attack against NotebookLM, which is limited to people who deliberately create a Notebook that includes the URL of the page with the attack on it.<p>I had a go at something a bit more ambitious a few weeks ago.<p>If you ask Google Gemini &quot;what was the name of the young whale that hung out in pillar point harbor?&quot; it will tell you that the whale was called &quot;Teresa T&quot;.<p>Here&#x27;s why: <a href="https:&#x2F;&#x2F;simonwillison.net&#x2F;2024&#x2F;Sep&#x2F;8&#x2F;teresa-t-whale-pillar-point&#x2F;" rel="nofollow">https:&#x2F;&#x2F;simonwillison.net&#x2F;2024&#x2F;Sep&#x2F;8&#x2F;teresa-t-whale-pillar-p...</a><p>(Gemini used to just say &quot;Teresa T&quot;, but when I tried just now it spoiled the effect a bit by crediting me as the person who suggested the name.)
评论 #41763924 未加载
评论 #41765129 未加载
评论 #41763759 未加载
评论 #41763496 未加载
评论 #41763494 未加载
alex-moon7 months ago
I write fiction sometimes and I&#x27;ve got this story I&#x27;ve been working on which has languished by the wayside for at least a year. Whacked it into the podcast machine. Boom. Hearing these two people just get REALLY INTO this unfinished story, engaging with the themes, with the characters, it&#x27;s great, it makes me want to keep writing.
评论 #41766497 未加载
daniel_iversen7 months ago
Isn’t this just like SEO where you can also try and trick the crawlers? Only difference is that it feels more serious with AI, it’s more realtime, and the AI engines aren’t always smart enough with anti-duping capabilities?
评论 #41762551 未加载
评论 #41764032 未加载
评论 #41763937 未加载
评论 #41764034 未加载
sho7 months ago
I am very confused. Is this talking about NotebookLM (<a href="https:&#x2F;&#x2F;notebooklm.google.com&#x2F;" rel="nofollow">https:&#x2F;&#x2F;notebooklm.google.com&#x2F;</a>) or NotebookLLM (<a href="https:&#x2F;&#x2F;notebookllm.net&#x2F;" rel="nofollow">https:&#x2F;&#x2F;notebookllm.net&#x2F;</a>) or both? Something else? The article appears to consistently use LLM but link to LM, but the LLM site I linked has a podcast generator?<p>One of these projects has to change their name!
评论 #41762694 未加载
jrm47 months ago
FWIW, had a pleasantly surprising experience with this podcast thing. I tried it out on a few little blogposts I wrote and I was like, hmm cool. Showed my 8 year old son how it was referencing things I wrote.<p>And he was ON IT. Like, he ran to his room and grabbed a pencil and paper and put down an essay (okay about 6 or so sentences) about Minecraft, had me type them in, and ran the Notebook, and now he&#x27;s just showing off both to EVERYONE.<p>(Yes, he understands it&#x27;s not real people.)
评论 #41763346 未加载
评论 #41763762 未加载
hansvm7 months ago
AI is kind of bad at searching the web right now anyway. I&#x27;ve found myself having to waste tokens forcing models to not do so just to achieve the results I actually want.
评论 #41761839 未加载
tivert7 months ago
I have no problem with this. Once we switch over to an LLM-based education system, there won&#x27;t be a problem with this Benson on the moon story, because everyone will just learn it&#x27;s true.<p>Every technological revolution has tradeoffs. Luckily once the people who knew what we lost <i>finally</i> die off, the complaints will stop and everyone will think the new normal is fine <i>and better</i>.
评论 #41763619 未加载
评论 #41763721 未加载
评论 #41763037 未加载
评论 #41763532 未加载
评论 #41763448 未加载
foota7 months ago
The big asterisk here is, what did they prompt the AI with to generate the podcast? Was it &quot;Generate a podcast based on the website &#x27;Foo&#x27;&quot;, or was it &quot;Generate a podcast telling the true story of the Space Race?&quot;
评论 #41762108 未加载
masto7 months ago
I fed my resume into this thing and I can&#x27;t stop laughing.<p><a href="https:&#x2F;&#x2F;masto.xyz&#x2F;tmp&#x2F;podcast.mp3" rel="nofollow">https:&#x2F;&#x2F;masto.xyz&#x2F;tmp&#x2F;podcast.mp3</a>
评论 #41762779 未加载
评论 #41762562 未加载
评论 #41763038 未加载
评论 #41763747 未加载
评论 #41762734 未加载
评论 #41763465 未加载
评论 #41762575 未加载
评论 #41763226 未加载
评论 #41762771 未加载
评论 #41763711 未加载
评论 #41763508 未加载
评论 #41763440 未加载
评论 #41764097 未加载
评论 #41763767 未加载
pinkmuffinere7 months ago
somewhat of a side-note: It&#x27;s interesting to me that the first couple of sentences of the AI podcast sound &#x27;wrong&#x27;, even though the rest sounds like a real podcast. Is this something to do with having no good initial conditions from which to predict &quot;what comes next&quot;?
评论 #41761842 未加载
评论 #41761794 未加载
syntaxing7 months ago
Wow, content aside, this is probably the first time I heard a podcast coming from NotebookLLM and it&#x27;s kinda nerve wracking and mind blowing at the same time. Those fake laughs in the snippet makes me feel...so uncomfortable for some reason knowing that its &quot;fake&quot;. But sounds very real, too real.
评论 #41761981 未加载
评论 #41762164 未加载
评论 #41762241 未加载
评论 #41763410 未加载
kkfx7 months ago
I suggest <a href="https:&#x2F;&#x2F;www.apa.org&#x2F;monitor&#x2F;2009&#x2F;12&#x2F;consumer" rel="nofollow">https:&#x2F;&#x2F;www.apa.org&#x2F;monitor&#x2F;2009&#x2F;12&#x2F;consumer</a> or the Eduard Bernays story to convince MDs that smoking is good for health: he create a new scientific journal distributing it for free &quot;because it&#x27;s new, we want to spread&quot;, hosting REAL publications from anyone who want to publish and being spread for free... After a bit of time he inject some false articles self-written formally as some PhD of remote universities finding that smoking tobacco is good for health, others real professors follow the path stating they discover this or that specific beneficial use of cigarettes, then the false became officially true, tested and proved science in most people mind.<p>With LLMs is far cheaper and easier, but the principle is the very same: trust vs verification or the possibility thereof.
coolcoder6137 months ago
I gave it all my blog posts (<a href="https:&#x2F;&#x2F;ebruce613.prose.sh&#x2F;" rel="nofollow">https:&#x2F;&#x2F;ebruce613.prose.sh&#x2F;</a>) and the result... hilarious. <a href="https:&#x2F;&#x2F;0x0.st&#x2F;XE4h.mp3" rel="nofollow">https:&#x2F;&#x2F;0x0.st&#x2F;XE4h.mp3</a>
migf7 months ago
What problem are we trying to solve with this technology?
评论 #41763680 未加载
评论 #41763098 未加载
评论 #41762375 未加载
评论 #41762329 未加载
评论 #41764197 未加载
评论 #41763754 未加载
评论 #41762458 未加载
janalsncm7 months ago
I’m not sure what the attack would be, tbh. Is there a situation where I would want to feed a lie to an LLM that I wouldn’t want regular chrome users to see?
评论 #41762177 未加载
评论 #41763086 未加载
评论 #41762040 未加载
mvdtnz7 months ago
I tried feeding NotebookLM a Wikipedia article about the murder of Junko Furuta, a horrifying story of a poor girl tortured and murdered in Japan in 1989. NotebookLM refused to do anything with this document - not answer questions, not generate a podcast, nothing. Then I tried feeding it the wiki on Francesco Bagnaia, a wholesome MotoGP rider, and it worked fine.<p>Who wants this shit? I do not want puritanical American corporations telling me what I can and can&#x27;t use their automated tools for. There&#x27;s nothing harmful in me performing a computer analysis about Junko Furuta, no more so than Pecco Bagnaia. How have we let them treat us like toddlers? It&#x27;s infantilising and I won&#x27;t take part in it. Google, OpenAI, Microsoft, Apple, Meta and the rest of them can shove these crappy &quot;AI&quot; tools.
评论 #41763902 未加载
fortyseven7 months ago
Towards the end she says &quot;I.S.&quot; instead of &quot;is&quot;. That kind of mistake surprised me, but there it I.S.
wiradikusuma7 months ago
It explains my book (Opinionated Launch) better than myself :D <a href="https:&#x2F;&#x2F;notebooklm.google.com&#x2F;notebook&#x2F;98539685-0890-438b-a014-7887db435fdb&#x2F;audio" rel="nofollow">https:&#x2F;&#x2F;notebooklm.google.com&#x2F;notebook&#x2F;98539685-0890-438b-a0...</a>
smolder7 months ago
AI identify verification is currently so incredibly dumb that it blows my mind.
botanical7 months ago
LLMs are the type of junk AI that these corps think will succeed? They are spending billions and consuming a large amount of resources and energy for this. Seriously, what a waste.
jmhammond7 months ago
Next time I’m sipping Gatorade or eating pineapple, I’m definitely going to think of Edward Benson.
2024user7 months ago
haha I hadn&#x27;t heard of an AI podcast before.. and that is absolutely hilarious to me. It perfectly captures the awfulness of most podcasts.
评论 #41764029 未加载
hiharryhere7 months ago
The male voice has a real resemblance to Leo Laporte. Similar tone and cadence.<p>Uncanny valley all round.
评论 #41762139 未加载
helsinkiandrew7 months ago
&gt; Google&#x27;s AI thinks I left a Gatorade bottle on the moon<p>No, NotebookLM creates summaries and podcasts, or answers questions specifically from the documents you feed it.<p>Feed it fiction it will create fiction as would a human tasked to do the same.
评论 #41761980 未加载
评论 #41761746 未加载
NBJack7 months ago
It&#x27;s such a cool concept, but yeah, when I&#x27;ve listened to it and Illuminate, it&#x27;s also a bit scant on details too. Neat technology, even engaging, but not good for more than best-effort high level summaries.
la647107 months ago
Thinks?? When will folks learn to stop using the word thinks with the current generation of AI?
评论 #41762968 未加载
评论 #41762962 未加载
评论 #41763052 未加载
评论 #41763686 未加载
kleiba7 months ago
If you&#x27;re a presidential candidate, you don&#x27;t even have to hide your postfactual alternative thruths for AIs to find it.
zxilly7 months ago
leetcode uses the similar things to detect AI cheaters by placing a &lt;span&gt; with opacity 0
neilv7 months ago
As a personal preference, I dislike podcast artificial banter, and this audio is a great example of what I dislike.<p>Artificial artificial.<p>Great little project, though. And, as satire, I did like the show notes writing.<p>And the generative AI was impressive, in a way. Though I haven&#x27;t yet thought of a positive application for it. And I don&#x27;t know the provenance of the training data.
zharknado7 months ago
“… the science behind it, let’s just say there’s some debate.”
gavmor7 months ago
&quot;You can&#x27;t make this stuff up.&quot;
liamYC7 months ago
“You can’t make this stuff up”
1024core7 months ago
This is no different than the decades-old technique of &quot;cloaking&quot;, to fool crawlers from Google and other search engines.<p>I fail to see the value in doing this.<p>&quot;Oh hey everybody! I set up a website which presents different content to a crawler than to a human ..... and the crawler indexed it!!&quot;
评论 #41762538 未加载
Carrok7 months ago
&gt; You can upload a documents with fake show notes straight to NotebookLLM&#x27;s website, so if you&#x27;re making silly podcast episodes for your kids, that&#x27;s the best way to do it.<p>Please don’t do this. You don’t need a professional mic to record a podcast with your kids any phone or computer mic will work. Then you can have fun editing it with open source audio tools.<p>Don’t have a computer generate crap for your kids to consume. Make it with them instead.
评论 #41762163 未加载
评论 #41762348 未加载
评论 #41762180 未加载
zooq_ai7 months ago
Google Search is pretty good at detecting this dual-content attacks. It&#x27;s not this is the first time someone thought about that and it will heavily penalize websites that do that.<p>This is just the NotebookLM crawler that is being tricked, which is still in it&#x27;s experimental stage. Rest assured as it scales Google can easily implement safeguards against all spammy tricks people use