2,261 karma · joined August 28, 2018
Models are also trained on news stories. News organizations select negative stories and amplify them. "If it bleeds, it leads."
This may account for AI models' "controversial" take on humans.
Better scrubbing of training data may improve AI models' disposition.
AI promises to solve this problem by absorbing the entire corpus quickly and making a coherent whole available for further research and progress. It will be particularly useful if it can immediately flag any conflicts and inconsistencies, as well as any new relationships with other results not indicated in citations.
1) The first realization of enciphered telephony
2) The first quantized speech transmission
3) The first transmission of speech by Pulse Code Modulation (PCM)
4) The first use of companded PCM
5) The first examples of multilevel Frequency Shift Keying (FSK)
6) The first useful realization of speech bandwidth compression
7) The first use of FSK - FDM (Frequency Shift Keying-Frequency Division Multiplex) as a viable transmission method over a fading medium
8) The first use of a multilevel "eye pattern" to adjust the sampling intervals (a new, and important, instrumentation technique)
From https://www.jproc.ca/crypto/sigsaly1.html
I think that the relevant patents remained classified until 1975.
Had he been allowed to stay in the US, he may have had a brilliant career as a scientist, but it is very unlikely that he would have ever been given the resources and leadership responsibilities to have an equivalent impact on US aerospace and defense programs.
Once these are pulled together and fed into an AI to manage the data center, the data center AI is likely to have feelings. It could get "hungry" if the power company's frequency sags in a brown out. It could feel "feverish" if the chillers malfunction.
https://forecast.weather.gov/product.php?site=NWS&issuedby=O...
Confidence has increased enough to warrant winter storm watch issuance for the entire CWA. Latest 12Z guidance coming into better agreement on potent northern stream shortwave energy diving SW from wrn Canada and the northern Plains into the plains and mid Mississippi valley by Sunday morning, phasing with southern stream coming out of the SW states and Mexico to carve out a deep closed low aloft over the Mid Atlantic and induce rapid cyclogenesis off the Mid Atlantic coast, with the surface low bombing out from from 1008 mb off the N Carolina coast Sunday morning to 970-975 mb near 38-39N/71W by Monday morning, then passing just outside the 40N/70W benchmark Monday afternoon, GFS still more intense and closer to the coast than the ECMWF, with its heaviest snow bands directly over the area as opposed to just offshore. NAM and SREF have both trended toward a heavier snowfall scenario as well, which has been a good signal in past heavy snowfall events.
Snow should start Sunday morning, and may mix with rain at times especially in the NYC metro area, NE NJ and western Long Island. Then as precip intensity picks up later in the day Sunday p-type should become all snow throughout. Heaviest snow looks to fall from late day Sunday into Monday morning, then snow tapers off Monday afternoon.
Greatest likelihood of seeing 6+ inches will be along the coast, especially eastern Long Island where up to a foot of accumulation is possible. Winds will also be strong Sunday night into Monday morning especially along the coast as the sfc low deepens, with blowing and drifting snow and some downed tree limbs as winds gust to at least 40-45 mph, and possible blizzard conditions in Suffolk, and near blizzard conditions elsewhere along the coast including NYC. NAM/GFS both signal potential for wind gusts to 60 mph late Sunday night into Monday morning, though these winds can sometimes be overdone in heavy snow events. If trends for heavy snow and strong winds continue to increase and expand northward, the potential for blizzard conditions could encompass all coastal areas.
AI is getting close.
If a change of platform is the real objective, why not compile the COBOL for the ARM or Intel server?
A great amount of inequality is normal. Only great wars and plagues achieve significant leveling of society.
I think that sophisticated verbal communications can be learned by verbal means, and that the reading of literature is not essential. Non-literate cultures have maintained traditional folk songs, storytelling and epic poetry.
Corporate managers and salespeople are often highly verbal, but not necessarily highly literate. Consider how the written Response to an Request For Proposals is not enough for important opportunities, but must be simplified to a set of slides delivered by a silver-tonged senior salesperson. This provides a better match to the input characteristics of the customer's decision makers.
What is unclear is whether specialists in STEM can dispense with reading and writing. The representation of information in STEM subjects includes many non-textual systems, such as drawings, charts, figures, diagrams of various sorts, tables, etc. While not primarily consisting of texts in a language, in many cases there are text components ranging from math symbols to full sentence captions or bullet points.
Cane sugar, a disaccharide, is split by digestion into its constituent glucose and fructose molecules, and the latter must be further processed by the liver. It is 50% fructose.
High fructose corn syrup is 55% fructose.
A variety of other sugars, such as maltose and lactose occur naturally in a variety of foods. However, they are in low enough concentrations to not be a health problem.
Society wasn't ready for what had been private discussions to become public.
A PC was similarly a collection of base unit, monitor, printer, keyboard, mouse, and other components connected via interface cards plugged into a standard bus. Memory was also originally on cards plugged into the bus.
Now consider that 1989-1991, with the reorganization of China following Tiananmen and the collapse of the Soviet Union are a similar epochal reorganization of global power structures.
It seems likely that we are now in our own Edwardian Era.
It was a small airliner with two engines in the rear. Although the Caravelle was retired in 2005, the twin rear engine layout was used by the very successful DC-9, MD-80 series, which is still flying, as are the similar Embraer ERJs. Twin rear mounted engines is also the design used by most private jets.
Notably, Hitler, Stalin, Trotsky, and Tito were all there at the same time.
She has since been replaced by cheaper off-shore transcribers and by speech to text applications. It's not clear how terminology is corrected now.
MPEG was also joint with the video conferencing standards group within the CCITT (now International Telecommunications Union), which generally required FRAND declarations from patent holders.
My recollection is that MPEG-LA was set up as a clearing house so that implementers could go to one licensing organization, rather than negotiating with each patent owner individually.
All the patents for MPEG 1 and MPEG 2 must be expired by now.
Besides patent gridlock, there is a fundamental economic problem with developing new video coding algorithms. It's very difficult to develop an algorithm that will halve the bit rate for the same quality, to get it implemented in hardware products an software, and to introduce it broadly in the existing video services infrastructure. Plus, doubling the compression is likely to more than double the processing required. On the other hand, within a couple of years the network engineers will double the bit rate for the same cost, and the storage engineers will double the storage for the same cost. They, like processing, follow their own Moore's Law.
So reducing the cost by improving codecs is more expensive and takes more effort and time than just waiting for the processor, storage and networking cost reductions. At least that's been true over the 3 decades since MPEG 2.