Internet

Internal records show that Twitter has shed about 80% of its employees since Elon Musk took over and headcount is hovering around 1,300 working employees today. With fewer than 550 full-time engineers now, one former Twitter engineer says the remaining team will be spread thin, and will likely have a hard time maintaining the service while adding new features. In addition to the 1,300 full-time Twitter employees, new owner and CEO Elon Musk has authorized about 130 people from his other companies, including Tesla, SpaceX and The Boring Co., to work for the social media business. Twitter’s full-time headcount has dwindled to approximately 1,300 active, working employees, including fewer than 550 full-time engineers by title, according to internal records viewed by CNBC. Around 75 of the company’s 1,300 employees are on leave including about 40 engineers. The company’s trust and safety team, which makes policy recommendations, design and product changes with the aim of keeping all of Twitter’s users safe, is down to fewer than 20 full-time employees. Internal records indicate that there are also about 1,400 non-working employees of Twitter who are still being paid, but are no longer expected to fulfill their old responsibilities at the social media company. Many of them resigned when CEO Elon Musk sent out a “pledge” asking them to commit to “hardcore” work at Twitter 2.0 including long hours.

via cnbc: Twitter is down to fewer than 550 full-time engineers

siehe auch: 80 percent of Twitter’s full-time staff has evaporated under Musk No wonder the site runs like trash these days. Elon Musk wasn’t lying last October when he told Bloomberg that 75 percent of the employees at his newly acquired toy, Twitter.com, wouldn’t lose their jobs under his ownership, as The Washington Post had reported at the time. Turns out, it’s closer to 80 percent. Of the roughly 7,500 people working there before Musk’s takeover, CNBC reports Friday that barely 1,300 in total, and fewer than 550 full-time engineers, are left at the husk of a company, either through said layoffs or voluntary resignations. CNBC also notes that 75 employees are currently on leave, 40 of which are engineers, while the Trust and Safety team, which oversees content moderation for the site, has been culled to fewer than 20 full-timers. This news comes at the end of a seemingly ceaseless string of blunders since Musk announced an unsolicited $44 billion bid to buy the social media site last April.

Internal records show that Twitter has shed about 80% of its employees since Elon Musk took over and headcount is hovering around 1,300 working employees today. With fewer than 550 full-time engineers now, one former Twitter engineer says the remaining team will be spread thin, and will likely have a hard time maintaining the service while adding new features. In addition to the 1,300 full-time Twitter employees, new owner and CEO Elon Musk has authorized about 130 people from his other companies, including Tesla, SpaceX and The Boring Co., to work for the social media business. Twitter’s full-time headcount has dwindled to approximately 1,300 active, working employees, including fewer than 550 full-time engineers by title, according to internal records viewed by CNBC. Around 75 of the company’s 1,300 employees are on leave including about 40 engineers. The company’s trust and safety team, which makes policy recommendations, design and product changes with the aim of keeping all of Twitter’s users safe, is down to fewer than 20 full-time employees. Internal records indicate that there are also about 1,400 non-working employees of Twitter who are still being paid, but are no longer expected to fulfill their old responsibilities at the social media company. Many of them resigned when CEO Elon Musk sent out a “pledge” asking them to commit to “hardcore” work at Twitter 2.0 including long hours.

via cnbc: Twitter is down to fewer than 550 full-time engineers

siehe auch: 80 percent of Twitter’s full-time staff has evaporated under Musk No wonder the site runs like trash these days. Elon Musk wasn’t lying last October when he told Bloomberg that 75 percent of the employees at his newly acquired toy, Twitter.com, wouldn’t lose their jobs under his ownership, as The Washington Post had reported at the time. Turns out, it’s closer to 80 percent. Of the roughly 7,500 people working there before Musk’s takeover, CNBC reports Friday that barely 1,300 in total, and fewer than 550 full-time engineers, are left at the husk of a company, either through said layoffs or voluntary resignations. CNBC also notes that 75 employees are currently on leave, 40 of which are engineers, while the Trust and Safety team, which oversees content moderation for the site, has been culled to fewer than 20 full-timers. This news comes at the end of a seemingly ceaseless string of blunders since Musk announced an unsolicited $44 billion bid to buy the social media site last April.

Für nicht einmal zwei Dollar die Stunde sollen diese der KI beigebracht haben, toxische Inhalte zu vermeiden – und mussten dafür genau diese sortieren. Es ist eines der zentralen Probleme bei der Entwicklung von künstlicher Intelligenz (KI): Wie verhindert man, dass diese all die Vorurteile, all den Hass, der im Internet zu finden ist, reproduziert? Dieses Problem des “Bias” stellt sich generell bei allen Formen des Maschinenlernens, bei großen Sprachmodellen wie dem Chat GPT zugrundeliegenden GPT-3 ist das aber besonders virulent. Eben weil hier weite Teile des Internets als Trainingsmaterial verwendet werden, eine Auswahl also nur begrenzt möglich ist. Bei Chat GPT heißt die Antwort darauf: Menschen. Und zwar Menschen, die der KI manuell beibringen, was eine passende Antwort ist und was nicht. So lernt diese nach und nach, was passendes Verhalten ist, tatsächlich neigt der Chatbot von Open AI im Gegensatz zu vielen seiner Vorgängern nicht zu rassistischen oder sexistischen Ausfällen. Das ist erfreulich, hat aber auch eine dunkle Seite, die nun ein Bericht von “Time” offenlegt. Das Training von Chat GPT soll nämlich von der kenianischen Firma Sama vorgenommen worden sein, deren Mitarbeiter dafür dann weniger als zwei Dollar pro Stunde bekommen haben sollen. Vor allem aber sollen sie dafür die dunkelsten Seiten der menschlichen Natur präsentiert bekommen haben. Problematische Inhalte Im Fall von Chat GPT soll das Training nämlich so abgelaufen sein, dass der KI beigebracht wurde, was toxische Inhalte sind. Konkret sieht das dann so aus, dass die Mitarbeiter all diese Inhalte manuell mit entsprechenden Labels versehen, mithilfe derer die KI dann nach und nach lernt, was sie aussortieren soll. Zu diesem Zweck sollen ab November 2021 zehntausende Negativbeispiele an die besagte Firma geschickt worden sein. Laut dem Bericht sollen darunter auch Beschreibungen von Mord, Folter oder auch sexualisierter Gewalt gegen Kinder gewesen sein. Stellungnahme Bei Open AI widerspricht man dieser Darstellung auch gar nicht, sieht all dies aber als essenziell in der Entwicklung von künstlicher Intelligenz an. “Die Klassifizierung und das Filtern von schädlichen Inhalten ist ein notwendiger Schritt zur Minimierung von gewalttätigen und sexuellen Inhalten in Trainingsdaten sowie bei der Entwicklung von Tools, die so etwas entdecken können”, heißt es in einer Stellungnahme.

via standard: Das schmutzige Geheimnis von Chat GPT sind kenianische Billiglohnkräfte

Von <a href=”//commons.wikimedia.org/w/index.php?title=User:Ni%C3%B0_ricsa%C3%B0&amp;action=edit&amp;redlink=1″ class=”new” title=”User:Nið ricsað (page does not exist)”>Nið ricsað</a> – <span class=”int-own-work” lang=”de”>Eigenes Werk</span>, CC BY-SA 4.0, Link