Challenges to Scaling AI
Businesses report that investments of any size into AI are paying off almost immediately. Those who believe in that might have skipped the long-term pilot projects before AI scaling. However, once businesses decide to operationalize AI and scale it up, that’s where the real problem begins. Although AI adoption is not limited to companies with the deepest pockets, like Amazon, Google, and Microsoft, it’s still quite expensive.
Here are some of the financial hurdles that businesses encounter when scaling AI:
Developmental Costs
For obvious reasons, businesses do not use third-party AIs when AI solutions require access to sensitive corporate data. So, they’ll have to build everything from scratch, which isn’t easy. Depending on the complexity of the AI, it may take a small team and a few months or a large team burning midnight oil for years before the AI achieves operational viability.
Data Costs
AIs are only as good as the data on which they are trained. However, high-quality data is challenging to obtain. Firstly, the data must be obtained from a highly reliable source to ensure accuracy. There are numerous pitfalls when sourcing training data for AIs, and it’s not always possible to identify or eliminate them.
Additionally, the data must be cleaned and prepared to make it easy for the AI to understand. Lastly, someone must secure the data fed into the AI to prevent loss or theft. Even Microsoft has had trouble with the last part!
Infrastructure Costs
AI systems require specialized hardware — usually GPUs — which can be significant cost centers. Organizations can keep operational costs low initially by using third-party cloud services and pay-per-use, but eventually, they may have to go in-house to achieve the correct scale and other business objectives.
Skill Costs
With the expansion in infrastructure comes the need for specialized AI and ML talent, which complicates the situation. AI-related jobs are among the hardest to fill globally. Besides the meteoric costs of hiring and training specialized AI talent, organizations must also implement training programs for others within the organization to help them use and contribute to the AI project meaningfully.
Costs of Competition
Last year, an industry expert estimated that OpenAI was likely spending up to $700,000 daily to keep ChartGPT-3 running whether to keep up with rapidly increasing competition, such as Google Bard and later Gemini, or as per their previously drawn-up strategy, OpenAI released ChatGPT-4, which rumoured to cost even more to operate.
As AI makers compete against each other to develop superior models, a kind of “arms race” among them sparks. And losing the race means handing over precious customers, revenues, market share, and the entire business to the competition. Then again, not participating in this race also means obsolescence for organizations.
With scalability riddled with so many challenges, we now have to confront the obvious question: is there enough evidence that AI is really living up to all the hype and delivering value at scale?
— Christina shares candid insights and ideas based on her work, network, and passion for mobile, payments, and commerce. As a frequent speaker by invitation to international events, from entrepreneurial and educational to executive audiences and settings, she has been recognized as a ‘Top B2B Influencer’, ‘Who’s Who in Fintech’, and ’40 Under 40 in Silicon Valley’. She focuses on the latest product innovations and growth for people during the day while teaching students and mentoring entrepreneurs at night. Connect with her on LinkedIn or Twitter/X. All views are my own. —
