HAL 9000 has arrived!

Aerospace engineering is very different from ordinary consumer engineering. Anyone who would put code like that in control of a life support system with no manual override wouldn’t get very far.

 

Boeing did with their 737 Max, two fatal crashes, and nearly did with their new spacecraft that could not reach the Space Lab.

Bob Wilson

 

My older brother got his physics masters at U of I Champaign in the 60's. I remember walking the halls of the Labs after attending one of his classes. The two memories that stand out are the illiac behind glass with a description from my brother of "this machine needed a minimum two man crew 24/7/365 available just to change vacuum tubes, and a glimpse of a tech in one of the labs working with a corded light pen and a CRT screen, drawing on the CRT with the pen. It took quite a few years after seeing that before I actually understood what I'd actually seen being done in the late 60's. Maybe someone else knows what the difference was between the eniac and the illiac from the 50's
U of I Champaign-Urbana

 

Based upon local reporting around Boeingtown, I understand that it did have a manual override, but not properly described in the operator manual and training materials.

MCAS itself wasn't disclosed to the pilots at all. It could be turned off, but the switches worked different than for the prior system, so that one of the "off" states on the old system was not completely "off" on the new. And a portion of recovery procedures necessary for this particular form of upset, included in the old-era 737 operator manual and training, had been dropped from more recent versions as apparently no longer needed, so today's generations of pilots were no longer aware of it. But this MCAS upset made it necessary again.

There were numerous other problems too. While grounded, the MAX received a far deeper dive into its avionics than had ever been intended, finding many other potential faults and forcing a serious overhaul with many changes.

 

If the DC-3 is the 'Tin Lizzie" of the airline world, then the The 737 is the VeDubbaya, and this platform has not been without it's little glitches along the way. To continue the analogy, the A320 will probably be the Corolla, benefitting by population density, modernity and hull numbers.
Still....the 737 (first flown a year before crewed Apollo missions) is about as far removed from the HAL9000 as one can be in the aerospace biz.
They still have a stick.
NOT a paddle.
Boeing has a lot of problems, and their reputation has suffered greatly and justifiably since the days when people used to say "If it ain't Boeing, then I ain't going!"
That having been said, I'd fly a G4-737 today!
...and tomorrow.
AND daily.

If you want to plaster the HAL-9000 label on an aircraft, I'd consider the A330.
ONE look at the cockpit (and Air France 447) will tell you everything you need to know......

Automobile drivers take heed.
Your steering wheel might be next!!!

 

I should have added that the other potential faults found in the deep dive inspections and tests were mostly not new items introduced by the MAX, but old ones long present in prior generations too. For things that would happen so very rarely that we just hadn't seen them happen yet. If fixed properly, without introducing new bugs, then the MAX series should be significantly safer than the earlier generations, but crashes are already so infrequent that it should take a long time for the differences to appear in the safety statistics.

Many of these items persisted for such a long time because the systems had a long track record of very good overall reliability. Aviation people are often reluctant to touch and improve such 'proven' systems for fear of introducing serious new bugs into systems that are already very good, a phenomenon that software people are very painfully familiar with.

 

I do not think it is accurate to call MCAS an AI system. It is standard industrial automation.