Shouldn't the biggest concern be that OpenAI either doesn't know about these breaches or is concealing their knowledge of them? I mean, as of yesterday their primary message on this track is "most aligned model yet".
A simple prompt that still stumps frontier LLMs most of the time is “create a pinball game”. They’ll put all the right pieces there but then fail to arrange them such that the game is truly playable. They’ll put a wall in the way of the launch chute so the ball can’t be launched. Or the flippers will pivot the wrong way. Or there will be holes such that the ball drops off the bottom without getting within reach of the flippers, etc.
Opus 5 is the first I’ve seen to “one shot” it (in a harness, so it was more than one LLM call).
Failed demos like this give weight to the argument that AIs need more of a world model, an understanding of how physics works to avoid obvious stumbles like this.
No tool-calling LLM is really "one shot" anymore, so we can easily repurpose that expression to mean "without further human feedback". The generated pinball games still seem to end up coming out broken. There's not much conversation about a good arrangement of pinball playing fields, and the LLM is a next token predictor. Pick any out-of-mainstream topic and same happens. Pelicans on bicycles used to be out-of-mainstream.
As a kid I didn’t understand what the 555 timer chip on the Apple II disk controller was doing but I learned the hard way that when you misalign the pins on the drive connector cable and the 555 chip releases its blue smoke you can’t use the drive anymore :(
I have read as well that the 555 was used in the game paddles for the Apple II. 555 + potentiometer (the part you turned) varied the length (duty cycle?) of a square wave which the Apple II used to determine the paddle position.
The Apple II family did indeed use 555 timers, in either 558 or 556 chips, to drive the timing circuit used to read paddle and joystick positions. The following article explains both the circuit and the reading code:
The port that was standardized on for PC joysticks was the dumbest possible one:
The joystick itself just had 1 potentiometer per axis, wired directly to the port. The port had no A/D, no timer, and no interrupt. Instead there was a GPIO and a capacitor. You discharged the capacitor with a GPIO write, and then polled the GPIO to measure when the capacitor was charged again. The number of iterations through your polling loop would be proportional to the position of the axis.
This is a pain to emulate if you aren't doing cycle-accurate emulation. IIRC Dosbox has a bunch of kludges and still doesn't get the joystick right for every game.
[edit]
To clarify the game port used a 558 (quad stripped-down version of a 555) as a schmitt trigger, so it generated pulses of a width proportional to the potentiometer position. I looked up the Apple II interface and it looks very similar, but with the caveat that accelerated versions (e.g. the IIgs) would always clock to 1MHz when reading the joystick port, compared to the PC that could run at a huge range of clocks (and CPI) over the lifetime of the port.
I remember using similar trick to use LEDs to sense light. Basically, charge the (reverse biased) LED capacitance, then measure how long it takes to discharge.
The lil circuit I had was LED bar, so I used it to sense finger position using that (other leds providing light, LED doing the sensing judging that light and comparing to rest
I hear it a lot but this gambling analogy breaks when you look at actual outcomes. If you went to Vegas and after a few pulls on a one-armed bandit could _reliably_ walk away with the jackpot we wouldn’t even call it gambling anymore.
reply