086351-004 SNAT604IFS BB 41948 提供可解釋的人工智能解決方案

該系統(tǒng)為操作人員提供最相關(guān)的數(shù)據(jù)，并在考慮各種權(quán)衡的基礎(chǔ)上推薦最有效的措施。它還跟蹤這些行動在實現(xiàn)其目標方面的有效性，突出未采取行動的情況，以便解決執(zhí)行障礙或沖突目標。這種透明性對于發(fā)現(xiàn)可以改進性能的實例非常有價值。最終，該解決方案以數(shù)字方式克隆最佳運營商，編纂工程最佳實踐，并優(yōu)化實時性能以提高盈利能力。

Rittenberg指出，實施是一個反復(fù)的過程，她強調(diào)了公民發(fā)展的重要性。在工廠工作的人，如最好的操作員或工程師，可以合作并建立在知識的基礎(chǔ)上，這有助于捕捉后端的機會并重復(fù)利用它們。雖然不可能第一次100%正確地獲得數(shù)字化知識，但迭代過程允許隨著時間的推移進行改進。Rittenberg提供了一個超越極限人工智能如何幫助一個復(fù)雜的大型煉油廠進行大量變革的例子。通過專注于一個特定的系統(tǒng)并迭代幾個月，精煉廠持續(xù)達到其商業(yè)計劃，并在頭兩個月提高了17%以上。

Rittenberg承認，有些人在實現(xiàn)像AI這樣的大型軟件產(chǎn)品時會不知所措，但迭代過程允許組織捕捉機會并隨著時間的推移而改進。

Frazer補充說，隨著組織采用增量方法，信任會隨著時間的推移而建立，這對于建立技術(shù)和員工之間的關(guān)系非常重要。信任在人工智能中尤其是一個難題，因為能源和工業(yè)領(lǐng)域的專業(yè)人士擔心依賴作為黑匣子運行的機器產(chǎn)生的結(jié)果——缺乏對如何獲得這些結(jié)果的透明度。風險太大了。

Rittenberg回應(yīng)說:“這就是認知人工智能——超越極限產(chǎn)品的一個關(guān)鍵原則——增加價值的地方，它通過提供可解釋的人工智能解決方案，并為人類提供清晰的審計跟蹤，以便于理解和一致采用。這也意味著，能源等高度監(jiān)管的行業(yè)，在最苛刻的環(huán)境中運營，并使用高度敏感的數(shù)據(jù)，將能夠信任認知人工智能來優(yōu)化其當前的運營。"

總之，超越極限人工智能幫助組織實時數(shù)字化知識和優(yōu)化運營。該公司的方法涉及迭代過程、協(xié)作和公民發(fā)展，允許組織捕捉機會并隨著時間的推移而改進。通過數(shù)字化知識，組織可以有一個真實的來源，并專注于一線員工要交付的最重要的東西。增量方法在技術(shù)和工人之間建立信任，使其成為實施的一個重要方面

Rittenberg continued, when it comes to implementation, Beyond Limits AI typically starts with a system to show how the organization can deliver better decisions in the moment across the entire system. Even in a plant, there are silos—and different operating units, maintenance, planners and schedulers. There are so many decisions being made that depend on access to the right data and a swift, thorough interpretation of that data. Beyond Limits’ cognitive reasoning technology helps support decision-making where data might be difficult to interpret–and where it can be daunting to choose from many recommendations. Beyond Limits homes in on the place that impacts fuels production—right at the interface between engineers and operators in the plants.

In the campaign plan execution workflow, operators are given objectives to satisfy, and they are also asked to choose from among a set of options. Given what they know, and the sea of data presented at them, they must decide what actions will best achieve their objectives at that moment in time. Varying individuals will make choices differently, and success will depend on many factors, including familiarity with the equipment, response time, understanding of priorities, and individual style/approach to problem-solving, just to name a few. By digitizing knowledge, the organizations can have one source of truth, allowing them to focus on the most important thing for the frontline worker to deliver.