Open Password – Donnerstag, den 1. März 2019

# 520

 

Project Consult – Newsletter – Ulrich Kampffmeyer – Silvia Kunze Kirschner – Informationsmanagement – Dow Jones – Open Password – DNA – US-Informationsindustrie – Artificial Intelligence – James Bell – Machine Learning – Dow Jones – 4th Industrial Revolution – Trust – Alpha Go – Google Deep Mind – Lee Seedol – God´s Touch – Bias – Training – Microsoft – Tay – Standards – Optical Disc – Format Wars – Toshiba – Sony – Browser Wars – Google – IBM – Ethical Governance – Donald Trump – Artificial Intelligence – Thomson Reuters – CLEAR Risk Information – EBSCO – Stacks – Dun & Bradstreet – Anthony Jabbour – SIX – Blockchain – SIX Digital Exchange – ALPSP – Plan S – Outsell – EBSCO – FH Potsdam – Information Literacy – Data Literacy – Stephan Büttner – Antje Michel

Briefe

Project Consult: Newsletter feiert
20. Geburtstag und kommt fast
auf 200 Ausgaben

Informationsmanagement
in den letzten 20 Jahren

Sehr geehrte Damen und Herren, 

es ist ein besonderer Newsletter-Tag für uns: der Newsletter hat im Februar 2019 seinen 20. Geburtstag gefeiert – und seine 198ste Ausgabe!

Viele wichtige Themen wurden und werden behandelt, viele Informationen diskutiert und hinterfragt, immer wieder auf die wichtige Weiterbildung und Qualifizierung für Enterprise Content Management/Dokumentenmanagement/Projektmanagement/Rechtsfragen usw.  hingewiesen …

Wir freuen uns, wenn Sie die Informationen aus unserem Newsletter für Beiträge und Event-/Seminar-/Veranstaltungshinweise nutzen können – und wenn ja, auf Ihren Beleglink.

Es bleibt spannend im Informationsmanagement – Soeben ist der PROJECT-CONSULT-Newsletter unter anderem mit diesen Inhalten erschienen (siehe Anlage-PDF oderhttp://bit.ly/PCNL-Jubi20Jahre):

Ulrich Kampffmeyer, Silvia Kunze Kirschner: “We wrote, we shared, we changed, we focussed“ … der PROJECT CONSULT Newsletter zum Information Management im Lauf der letzten zwei Jahrzehnte

Dietmar Weiss: „Elektronische Rechnungsbearbeitung: Wenn nicht jetzt, wann dann?“

Ulrich Kampffmeyer: „SICS – selbstinstallierende und selbstkonfigurierende Systeme“

Ute Lafrenz: „DSGVO-Vorgaben effizienter erfüllen?“

Ulrich Kampffmeyer: „Konsistentes Löschen in der Block Chain“

Mit freundlichen Grüßen

Silvia Kunze-Kirschner, PROJECT CONSULT Unternehmensberatung Silvia Kunze-Kirschner, HamburgHamb

Unternehmen des Jahres

Dow Jones als Repräsentant
der US-Informationsindustrie

Open Password hat Dow Jones als Repräsentant der US-Informationsbranche zum Unternehmen des Jahres gewählt. Die Bedeutung von KI für unsere Branche und die These, dass die Künstliche Intelligenz auch in Deutschland vor dem Durchbruch steht, wurde an anderer Stelle begründet („Vor dem Durchbruch im deutschsprachigen Raum“, #491, 8. Januar). Dow Jones hat weltweit und im deutschsprachigen Raum Wesentliches zum Vormarsch der US-Informationsindustrie in die Artificial-Intelligence-Welt beigetragen, mit grundlegenden Papieren, mit dem Kolloquium „The Art of the Possible: Creatively Advancing Data Science Across the Enterprise“ im Dezember in Frankfurt und mit seiner Plattform DNA, die den Unternehmen Daten für die Entwicklung eigener Anwendungen für Künstliche Intelligenz anbietet. Dem heutigen Beitrag von James Bell – „Trust in the Revolution“ – werden weitere Beiträge von Dow Jones zur Künstlichen Intelligenz folgen.

Wer den Nachrichtenteil von Open Password aufmerksam verfolgt hat, weiß, dass sich Dow Jones die Auszeichnung auch als Repräsentanten der gesamten US-Informationsindustrie verdient hat. Hier kann man manchmal den Eindruck gewinnen, dass sich die Product Releases auf der Basis von AI geradezu überschlagen. Ein weiterer in KI-Angelegenheiten sehr rühriger Informationsanbieter und Wettbewerber von Dow Jones, der gleichfalls weltweit präsent ist, ist LexisNexis, dessen Grundsatzpapier „The Rise of AI and How it Will Impact Communication Professionals“ wir soeben veröffentlich haben (#511, 12. Februar).

Ein wichtiges Thema ist, wie die deutschen Anbieter und Nutzer auf die Herausforderung aus Übersee reagieren. Open Password möchte ihren Weg in die Anwendungen Künstlicher Intelligenz publizistisch begleiten.

Unternehmen des Jahres:
Dow Jones

Trust in the Revolution

By James Bell, Head of AI and Machine Learning, Professional Information Business, Dow Jones

“We should consider an aspect of AI developme that is frequently ignored outside the realms of certain intellectuals and specialists.”

For some, the rise of AI technologies has been likened to the dawning of a 4th Industrial Revolution. For others it is perhaps best described as a gathering storm; something that will blow over. However, this is to miss two important things.

The first is that the rise of AI is predominantly a human-made phenomenon that has been coming for a long time; slipping into our lives gently and slowly until we depend on its predictive analysis in our everyday lives. In other words, it is way too late to batten down the hatches, and it is not going to blow over. AI is here to stay.

The second thing to consider is that industrial revolutions are where a society chooses “who it wants to be”. That is the choice upon us now, and with it comes the knowledge that AI led analytics, when left unchecked, is a potent recipe for chaos. No further evidence needs mentioning than the impact upon recent elections and referendums of the very targeted advertisements on social media, which were fed by analytics and powered by AI algorithms.

So, perhaps we should consider an aspect of AI development that is frequently ignored outside the realms of certain intellectuals and specialists.

Trust.

First, trust in the aspect of quality. Can you trust in your Machine Learning algorithm’s quality? While tools and methods exist to assess accuracy and performance technically, how can a business manager trust that the resulting model is not about to lead your business astray? For even the most powerful can be disrupted. Consider the famous AlphaGo board game playing algorithm from Google DeepMind, which issued a challenge to the 9th Dan Professional Lee Sedol for a best-of-5 match at „Go“. At this event, watched by 60 million people, the startled Lee lost the first three matches in a row – a machine achievement previously thought impossible.

Then, in game four, Lee did something the AI couldn’t comprehend. Move 78 was a stone placed in the middle of the board; but, in a position so unlikely that the odds were calculated to be 10,000 to 1. It became known as the „God’s Touch“ because it perplexed and disrupted the algorithm for the rest of the game as Lee prevailed. There are many lessons from AlphaGo’s match against Lee Sedol, but perhaps the most important is highlighting that even the most powerful systems are only narrowly intelligent and at the mercy of their design, coding and training data.

So, how can one assess the quality of an algorithm making decisions in place of a human? This assessment requires, at a minimum, some level of transparency – a troublesome prospect in the current market as algorithms are locked away from prying eyes in the black boxes of “proprietary intellectual property”, a label that one Data Scientist described to me as, “probably an excuse to use linear regression”. Here the trained model’s code, its obtained weights and configurations, are acting as the modern equivalent of a Chinese Speaking Room, the thought experiment where information is only exchanged through a slot. For a business to be able to trust AI development and even more so in regulated markets such as Financial Compliance, someone must be able to peer through that letterbox and see what is going on inside to produce a chain of accountability that doesn’t end at the AI, instead it starts with it.

The second aspect of trust is the border between public and private data. Information over the border tends to be contentious and open to all sorts of bias. For example, a recently reported Google memo where engineers discussed putting their thumb on the scales of the search results to: “Actively counter prejudiced, algorithmically biased search results from search terms ‘Mexico’, ‘Hispanic’, ‘Latino’, etc.”

How can a client trust that the company processing corporate and private data with machine learning isn’t reproducing bias? Machine Learning and Neural Networks are often trained from data sets of historical human decisions and therefore can mirror the human bias inherent in those data sets. In other words, if the human decisions contained a conscious (or unconscious) bias, then the resulting model will do too.

Probably the most famous example of this was when Microsoft introduced the Twitter bot „Tay“, which was trained on data to model the friendly tweets of a 14-year-old valley girl. Unfortunately, Tay was using some of the tweets directed back at it to retrain the bot’s topics. This led Tay, in an astonishingly quick time period, to tweeting out racist, transphobic and fascist messages over 96,000 tweets in 15 hours before Microsoft was able to pull the plug.

The response to this is to demand of our developers, and those setting their task statements, a level of security around private data and create stronger methods of mitigating or even removing bias. Otherwise, how can we trust that our AI-controlled devices are not making decisions on data garnered from inappropriate sources?

The final consideration for trust is one of standards and governance. The value of common standards is that they enable a maturing of solutions and the much wider adoption of ideas. The problem is that new markets, such as those for AI, are the wild west for standards, which increases costs. For a historical example with business impact, consider the High-definition optical disc format wars of 2006 to 2008. Movie producers had the difficult decision to make of supporting either the Toshiba supported HD-DVD format or that of Sony’s Blue-ray. Either that or face the extra cost of releasing on both. This format race analogy applies equally to the internet age, where early giants like Microsoft released proprietary software „extensions“ for their web servers in an attempt to dominate the “browser wars”.

The same things can be seen in AI technologies, where the need for unified standards are becoming ever apparent, and standards are always led by governance and best-practice. How can you as a company that is considering adopting AI technology be sure that it has been created with the best standards of security, governance and ethics? No universal standard for these things currently exists, and every company is attempting to dominate the market through their various strategies, be it through sharing their tools (Google) or by ring-fencing their comprehensive offerings behind a brand paywall (IBM).

Thankfully, AI trust is starting to be taken seriously. As these ethical governance frameworks develop, the standard battles will crown a champion and we will hopefully discover new ways that AI and humans can live alongside each other. As large organisations and data stewards consider their future applications and capabilities, perhaps trust will become a new form of currency in the AI-led future. By building trusted automation technology that is transparent, ethical and augments humanity, perhaps companies will give us all some sort of umbrella to protect us from what could lie ahead.

For it is definitely looking like rain on the horizon.

Trump für eine Förderung
intelligenter Lösungen

President Trump signed an executive order meant to spur the development and regulation of artificial intelligence, technology that many experts believe will define the future of everything from consumer products to health care to warfare. A.I. experts across industry, academia and government have long called on the Trump administration to make the development of artificial intelligence a major priority. Now, Mr. Trump has taken that step, though this “American A.I. Initiative” might not be as bold as some had hoped.

Thomson Reuters mit Risikoeinschätzung auf Basis öffentlicher Daten.  Thomson Reuters has launched CLEAR Risk Inform, a new product to improve the risk-analysis process for organizations and the first risk-scoring solution to bring together analysis and public records with customizable definitions. CLEAR Risk Inform allows organizations to create an institutionalized risk assessment from public data.

EBSCO übernimmt Stacks. EBSCO Information Services (EBSCO) gibt die Übernahme von Stacks Inc. bekannt, dem Anbieter einer Webplattform für Bibliotheken. Im Jahr 2016 ging EBSCO mit dem Unternehmen Stacks Inc. eine Partnerschaft ein, um deren gehostetes Content-Management-System für Bibliotheken auf den Markt zu bringen. Die Übernahme zeigt EBSCOs Bestreben, Bibliothekskunden die bestmögliche Sucherfahrung und vollständig integrierte Ressourcen zu bieten.

Übernahme von Dun & Bradstreet abgeschlossen. Dun & Bradstreet and an investor group led by CC Capital Partners, LLC, Cannae Holdings, Inc., Bilcar, LLC, Black Knight, Inc. and funds affiliated with Thomas H. Lee Partners, L.P., along with a group of other investors, announced the completion of the Investor Group’s previously announced acquisition of Dun & Bradstreet. In connection with the closing, Anthony Jabbour, Black Knight’s Chief Executive Officer, was appointed Chief Executive Officer of Dun & Bradstreet and will remain in his current role at Black Knight.

Handelsplattform mit Blockchain. Swiss exchange SIX plans to launch its new SDX trading platform using blockchain technology to speed up trading in the second half of this year. The new SIX Digital Exchange (SDX) will initially run parallel to the existing SIX platform, which involves three steps to complete a trade, often over several days. Two of those steps vanish in a blockchain distributed ledger, meaning a transaction can be completed in fractions of a second.

ALPSP mit Einwänden gegen jetzigen Plan S.ALPSP has submitted a formal response to the Call for Feedback on Guidance on the Implementation of Plan S. The ALPSP position statement is as follows: ALPSP has concerns regarding the pace of change advocated by Plan S; it requests more clarity regarding the requirements for transformative agreements with opportunity for further review; it advocates the retention of different licence options; funding is not universally available to those publishing research; and the global publishing landscape is complex.

Quelle: Outsell, EBSCO

Briefe: 8. I-Science-Tag
am 15. Juli

Daily Digital – Data & Information Literacy

Der Fachbereich Informationswissenchaften an der FH Potsdam lädt zum 8. I-Science-Tag „Daily Digital – Data & Information Literacy“‘ ein. Sie schreibt:

„… „Data Literacy“ und „Information Literacy“ werden somit zu Schlüsselqualifikationen für die Teilhabe an einer digitalisierten Lebens- und Arbeitswelt. Folgende Fragen bilden den informationsdidaktischen Fokus des diesjährigen I-Science Tags:
– Was sind Data Literacy und Information Literacy in einer digital geprägten Lebens- und Arbeitswelt und wie ist das Verhältnis zueinander?
– Für wen sind diese Schlüsselkompetenzen gesellschaftlich relevant und was für konkrete Anforderungen stellen unterschiedliche Wissenskulturen an diese Kompetenz-Sets?
– Welche informationsdidaktischen Vermittlungskonzepte sind für die Entwicklung von Data Literacy und Information Literacy konzeptionell angemessen und didaktisch praktikabel und welchen Stellenwert haben digitale Vermittlungskonzepte?

Die Tagung ist als interdisziplinäres Austauschformat konzipiert.  … Ansprechpartner: Prof. Dr. Stephan Büttner: st.buettner@fh-potsdam.de, Prof. Dr. Antje Michel: michel@fh-potsdam.de, Telefon: +49 (331) 580 1517.“

Anzeige

FAQ + Hilfe