blueboo

12 ms·

121.
by blueboo 3y ago
I used to be a cheerleader for trains in the US. But there are a dozen setbacks for every win. Building infrastructure like this just isnt in the cards at this stage overnight on land will happen on self driving cars before it happens on ra
122.
by blueboo 3y ago
Depends on your goals. People are beginning to shop primarily on TikTok and Instagram. Social media seems to be enshittify-evolving to be the new e-commerce. Meanwhile, summarising posts in video on Shorts/Reels/TikTok might be qu
123.
by blueboo 3y ago
And that will, in due time, get ads too, until you upgrade to Nitro+ And that’s after the servers start paying for the privilege of supporting some significant slice of those features…
124.
by blueboo 3y ago
This is the natural next stage for enshittification as Discord starts to claw back value from users. It will get worse before it gets even worse https://www.wired.com/story/tiktok-platforms-cory-doctorow/
125.
by blueboo 3y ago
If it looks like a bubble and sounds like a bubble… But AGI
126.
by blueboo 3y ago
This is the latest in a particular thread in Earth science research. Most prominently reminds me of the EU’s Destination Earth https://www.ecmwf.int/en/about/what-we-do/environmental-serv... - another digital
127.
by blueboo 3y ago
But the bad news, then, is all that complexity is hidden, metastasising
128.
by blueboo 3y ago
OpenAI's "end goal" has OpenAI becoming a >$1T (100x their >$1B investment) company en route to its non-profit incarnation. Hardly benign
129.
by blueboo 3y ago
Beware popularity in the Landsraad
130.
by blueboo 3y ago
The bitter lesson, my dude. http://www.incompleteideas.net/IncIdeas/BitterLesson.html If you find a simpler, trainable structure you might be onto something Attempts to get fancy tried and died
131.
by blueboo 3y ago
Classic generative art project. TouchOSC would’ve been a good fit
132.
by blueboo 3y ago
See Appendix D in https://arxiv.org/pdf/2303.10130.pdf OpenAI’s “GPTs are GPTs: An early look at the labor market impact potential of large language models” It’s not just construction workers. Still safe to be a pro at
133.
by blueboo 3y ago
I have mixed feelings associating “less freedom” with a commercial service. But it is overreaching to say there’s “no reason” to censor it. Friction of information discoverability matters in the aggregate. I do see your point that a dogged
134.
by blueboo 3y ago
Joining the chorus of “applications exist but functional agents don’t”. There is one proven application: raising credulous VC money—and hoping that funding lasts until someone else’s foundation model makes it work
135.
by blueboo 3y ago
It’s ok not to finish. Starting, discovering, untangling, solving root node problems can be rewarding and worthwhile. Not every moment’s labor has to feed into a SaaS launch. Follow problems that bother you and subjects that fascinate you.
136.
by blueboo 3y ago
This is an observed behaviour in large models, which tend towards “sycophancy” as they scale. https://www.anthropic.com/news/towards-understanding-sycopha...
137.
by blueboo 3y ago
“People don’t leave bad companies, they leave bad managers”, an aphorism that ringers truer with every departure I observe. (But I also see how the incentives of the org can erode at the energy put into cultivating management skills and pri
138.
by blueboo 3y ago
If it’s bargaining power, then it’s a power dynamic that may merit being called out, just like gaps in pay by other characteristics. Some of those characteristics were given protection in law after enough noise was made.
139.
by blueboo 3y ago
Approximate same work, approximate same cost of living, different salary. What’s not to resent?
140.
by blueboo 3y ago
It’s more like playing baseball with a whiffle ball or using a touchscreen over a mouse, a Segway over hiking. You can learn a lot about steering riding a bike with training wheels. But just like your kids, they’ll be physically unable to u
141.
by blueboo 3y ago
24 hours * 60 minutes * 5000 messages * 100 char/msg = 720mb. 14 years of unique messages would fit on 1gb sd, probably cheaper than a wifi unit and the interface required to configure it. it'll last longer too than the chatgpt ap
142.
by blueboo 3y ago
Similar challenges with computer keyboards, chessboards, musical instruments generally
143.
by blueboo 3y ago
Aww, the STT seems to fail with this error in console https://snipboard.io/qm6KuB.jpg interesting tho
144.
by blueboo 3y ago
Maybe you’re implying this — it’s an obvious omission — but his credibility should encourage us to read “character” and “story” for their evocative imagination and not, say, for fact.
145.
by blueboo 3y ago
There’s a grain of truth here, but I object to the hyperbolic headline as it is unsupported by the content’s described data points which are exceptional rather than representative of typical content. “The internet is broken in a fundamental
146.
by blueboo 3y ago
80/20 rule surely applies. Licensing the most prominent IP holders will get the vast majority of data secured. These billion-dollar-funded efforts can comfortably chase down, say, hundreds of deals. GPT-4 finished training g 18mo ago…p
147.
by blueboo 3y ago
Indicates Phi-3 and the next cohort will obsolete Phi-2
148.
by blueboo 3y ago
Creating an eBook that looks good on the current cohort of popular devices and platforms isn’t trivial. So there’s a cost to production (and a temporal aspect to shifting standards) There are non-trivial implications associating a digital
149.
by blueboo 3y ago
The elephant in the sub-tweeting room: OpenAI getting caught red-handed memorising many many NYT articles. Now, the root cause must be attacked: copyright itself, not the intransigence of OpenAI to license this content (as they have with, s
150.
by blueboo 3y ago
Conspicuous absence of evidence or even specifics of what the author thinks “sounding like ChatGPT”. I personally wouldn’t characterise it as “robotic” or akin to a trope-y robotic character (Jane from the Good Place). But credulously takin

More ›