Artificial intelligence needs to learn the lessons of digital transformation’s tendency to fail

In January this year, IBM released a detailed research report explaining that digital transformation can only provide a return on investment of -5% to 10%, rather than the expected 150%. This gap is related to the experience accumulated over the past few decades in client/server implementation, operating system migration, big data applications and technology implementation.

Not all technical implementations miss the mark, but most implementations do. The main problem is that the technology is immature, and technology providers and deployment companies often use sales pitches to attract buyers but lack follow-up to ensure that the promised value is delivered.

The same thing may happen with artificial intelligence.

When the client/server trend took off, the technology wasn't ready, leaving IBM in trouble and the market rushing to move into an area that wasn't yet mature.

Sales can usually go beyond the product itself, because sales often emphasize the added value and services of the product. In the era of the rise of new technologies such as artificial intelligence, many companies hope to gain a foothold in this field. However, apart from companies like IBM and Nvidia, which have been researching artificial intelligence for decades, no other company (including Google) has announced that it is ready to fully enter the field of artificial intelligence.

The reason IBM is so excited is because it owns WatsonX, one of the most mature artificial intelligence solutions on the market. In the field of enterprise-level generative AI, IBM is the most mature, while other companies rely on little or no basis in sales and marketing commitments to survive.

Buyers may suffer when sales are ahead of technology. Data shows that many people fail to do their due diligence, leading to this situation.

Solution: Do your homework and follow the process

A successful strategy in this situation is to adopt a “test first” approach. After ensuring that the solution provided by the supplier is mature and complete, its feasibility is verified through a pilot project. Even if the product is mature, it needs to be deployed gradually according to the actual situation to avoid possible large-scale failure. Failures in pilot projects are acceptable and correctable, leading to more informed decisions during the production phase.

Before piloting, make sure the vendor’s revenue and ROI requirements are achievable and seek references from companies that have successfully deployed the technology. Ask the vendor if the technology has been deployed internally, and talk to IT staff at companies using the technology to get real feedback.

Research and get best practices with others who are trying the same task, and be aware that not every solution will work for every company or even every department.

Hybrid multi-cloud is the practice of providing the best balance between uptime, cost, availability and reliability. It takes a vendor that understands this concept, has a deep relationship with a cloud provider you trust, and has gained enough experience that it shouldn't be learned on the job.

Especially for AI data, quality is critical and you need a lot of help to ensure it. You don’t want an AI that is biased or hallucinating, just like you don’t want an analysis that always provides inaccurate answers.

These new AI capabilities are expected to be multi-modal, including natural language, images, audio, video, and even the critical element of time. The use of AI often tends to optimize for one data type and perform poorly on others, so you need to understand the differences and let the vendor know that in areas where it is not capable, the other vendor Business may be a better choice.

Finally, you need help with metrics and milestones so that if a vendor is underperforming, you can identify the problem early and either switch vendors or teams. If the vendor you’re working with can’t help you set metrics and goals for your project, you’re working with the wrong vendor.

The problem we often encounter in recent big technologies, from client/server in the 1980s to artificial intelligence today, is that sales far outpace the product and support structure. The result is a deployment that fails to meet goals and expectations. In many cases, it’s wiser to wait until the right partner, the right team, and the right solution come along.

The above is the detailed content of Artificial intelligence needs to learn the lessons of digital transformation’s tendency to fail. For more information, please follow other related articles on the PHP Chinese website!