Kalev Leetaru Contributo

As deep learning has become ubiquitous, evaluations of its accuracy typically compare its performance against an idealized baseline of flawless human results that bear no resemblance to the actual human workflow those algorithms are being designed to replace. For example, the accuracy of real-time algorithmic speech recognition is frequently compared against human captioning produced in offline multi-coder reconciled environments and subjected to multiple reviews to generate flawless content that looks absolutely nothing like actual real-time human transcription. If we really wish to understand the usability of AI today we should be comparing it against the human workflows it is designed to replace, not an impossible vision of nonexistent human perfection.

While the press is filled with the latest superhuman exploits of bleeding-edge research AI systems besting humans at yet another task, the reality of production AI systems is far more mundane. Most commercial applications of deep learning can achieve higher accuracy than their human counterparts at some tasks and worse performance on others.

Instead of comparing deep learning algorithms against trained humans placed in the same situations, we as a society have developed the habit of comparing them against idealized but entirely non-existent utopian baselines that look nothing like actual human performance on those tasks.

Take the rapidly growing use of deep learning to understand human speech, driven by the desire for ever more accurate smart speakers, personal assistants and other voice interfaces. Today the performance of such systems is frequently compared against human baselines that were coded offline through multi-coder reconciled workflows designed to yield absolutely flawless transcriptions.

JUNE 24, 2019 BY DAVID EDWARDS

Technology progresses daily and with this progression come threats and risks to social, financial and economic life.

Today, cyber-attackers have resorted to the use of automation to launch more frequent attacks on different businesses and corporations.

While cyber-attackers are expending a lot of resources to launch more sophisticated attacks, many organizations still rely on manual efforts to gather internal security findings and contextualize them with external threat information.

It was reported that carelessness of employee was the reason behind the ransomware attack in 51 percent of the cases.

However, such outdated methods and strategies need to part way for AI because they use up a lot of time, in which cyber-attackers can successfully take advantage of vulnerabilities to breach systems and steal data.

Today, many organizations are looking towards the efficient and effective use of artificial intelligence (AI) to address cyber risks and control operations. Artificial intelligence is very crucial to cybersecurity because, with it, new vulnerabilities and weaknesses can be identified promptly, dealt with and analyzed to help prevent further attacks.

AI, because it is smarter and more efficient has the ability to reduce pressure on humans. In 2018, 845.37 million malware were created and around 10 million new malware are created every month this year.

Unsurprisingly, traditional cybersecurity methods are inefficient in handling such a huge variety of threats. Many organizations lack security professionals and so, there is a very huge challenge on the shoulders of organizations in dealing with big data sets in security.

And since cyber-attackers already make use of high-end machines and automation to increase the effectiveness and sophistication of their attacks, it is only proper for organizations to adopt AI in order to combat this problem.

En Catalunya, 179 empresas se dedican a esta tecnología que promete transformar industria y servicios

Hablar de inteligencia artificial en la empresa es como hacerlo del tiempo en un trayecto de ascensor. Esta tecnología tiene cada vez más presencia en la industria y los servicios de empresas de todo el mundo. Mediante algoritmos que imitan el razonamiento humano, la inteligencia artificial permite automatizar decisiones y predecir resultados a gran escala. Funciona en la nube e integrada con maquinaria. La utilizan tanto empresas de automoción y robótica, como bancos, consultoras o marcas de moda, por poner algunos ejemplos.

“La inteligencia artificial tiene un potencial enorme porque será un motor de transformación en todos los sectores”, comenta Daniel Marco, director general de Innovación y Economía Digital de la Generalitat. “Permite aumentar la productividad, ahorrar tiempo y costes de producción e incrementar los volúmenes de demanda y oferta”. En el 2030, se calcula que esta tecnología tendrá un impacto del 14% del PIB mundial, lo que en cifras se traduce en unos 15,7 billones de dólares. En el sur de Europa supondrá el 11,5% del PIB, unos 0,7 billones de dólares”, asegura Marco al valorar un estudio que hoy publica el Departament de Políticas Digitales y Acció, la agencia por la competitividad de la Empresa de la Generalitat.

A la cabeza: Estados Unidos y China son los países que lideran el desarrollo de esta tecnología

Par Anaïs Cherif 

Toutes les semaines, La Tribune braque les projecteurs sur une pépite méconnue de la French Tech. Cette semaine, AZmed. Lancée en août 2018, cette jeune pousse parisienne développe une intelligence artificielle pour aiguiller les radiologues et les urgentistes dans leurs diagnostics. L’algorithme est capable de détecter une anomalie en moins de 0,2 secondes.

Proposer un diagnostic fiable, tout en gagnant du temps. C'est l'équation à laquelle AZmed promet de répondre grâce à son intelligence artificielle capable de détecter les fractures et lésions osseuses sur les radios. Lancée en août 2018, la jeune pousse parisienne développe un algorithme de deep learning (apprentissage profond à l'aide de réseaux de neurones artificiels) qui s'intègre aux logiciels traditionnels utilisés par les radiologues et les urgentistes. « Lors de la lecture d'une radio, notre technologie détecte en moins de 0,2 secondes une anomalie et la signale par un rectangle », explique Julien Vidal, Pdg et co-fondateur d'AZmed.

La startup a été co-fondée par Alexandre Attia, diplômé d'un Master en intelligence artificielle spécialisé en imagerie médicale, de l'École Normale Supérieure Paris-Saclay, et Elie Zerbib, chirurgien au service obstétrique et gynécologie de l'AP-HP (Assistance Publique - Hôpitaux de Paris).

« Les fractures représentent la grande majorité des radios traumatologiques réalisées en France. Or, cet examen est souvent perçu comme rébarbatif et chronophage par la profession, tout en étant peu engageant d'un point de vue médical. C'est pourquoi notre technologie permet de gagner du temps, pour en dégager davantage avec le patient et réaliser des examens à forte valeur ajoutée », poursuit l'entrepreneur.